SlideShare uma empresa Scribd logo
1 de 7
Baixar para ler offline
Copyright 2015 Page 1 of 7
Addressing Security in SDN Environment: A Perspective
Saikat Chaudhuri
Tata Consultancy Services,
Bangalore
+91-8884742237
chaudhuri.saikat@tcs.com
Subash Adigal
Tata Consultancy Services,
Delhi
+91-9643330517
subash.adigal@tcs.com
ABSTRACT
For the last two years, there is a definite buzz around SDN which
is complemented by NFV. The strong focus on SDN is vindicated
by the fact that by 2016, there is a prediction that more than
10,000 enterprises will deploy SDN in their networks, which is
almost a ten-fold increase from the data in 2014 [1]. Also,
according to a survey done by Juniper in 2014, out of 400 U.S.
Businesses, around 52.5 percent of the companies plan to adopt
SDN in the future. 27 percent out of these companies have stated
that they are ready or almost ready to adopt SDN [2].
However, along with the future prospect of SDN, there are
numerous concerns surrounding the security aspect associated
with it. Despite the concerns, organizations are keen to adopt
SDN due to its numerous benefits amongst which analytics,
reporting and resulting automation are some of the key elements.
The White Paper discusses a possible solution towards addressing
the security concern by automating the security in SDN
environment. The solution deploys an application that
continuously monitors the alerts generated by IDS, analyses the
reports and automates action to block flows from hosts based on
the alerts raised by the IDS. The paper also discusses the
implementation of sFlow based monitoring of the network. This is
based on packet sampling which is utilized to analyze the traffic
trend and determine if there is a possible flooding attack which
may have gone undetected in the network.
Permission to make digital or hard copies of all or part of this work for
personal or classroom use is granted without fee provided that copies are
not made or distributed for profit or commercial advantage and that copies
bear this notice. To copy otherwise, or republish, to post on servers or to
redistribute to lists, requires prior specific permission and/or a fee.
TACTiCS – TCS Technical Architects’ Conference 2012
1. INTRODUCTION
1.1 SDN and NFV
SDN is a term coined to exhibit the concept in networking where
the control plane is decoupled from the physical infrastructure and
is instead moved out into a separate entity called the controller.
The controller provides networking services that help to manage
the network and also determine how the data plane flows are
configured. Since the controller runs as a separate software
application to manage the entire network, hence its termed as
‘Software Defined Networking’ aka SDN. The switches (physical
infrastructure), that handle the traffic flow, may be made available
from multiple vendors as long as they are compatible with the
protocol of communication between the controller and the switch.
This reduces dependency on any particular equipment
manufacturer and the service provider has the flexibility to use
switches of their choice that gives them the cost and/or
operational advantage.
NFV complements the concept of SDN by providing the
additional flexibility of dynamically providing network functions
on VMs. This enhances the levels of utilization of the physical
hardware and also bring in cost benefits where number of VMs
can be scaled up or down dynamically as per the requirement. The
VMs may be spawned on top of physical hosts such as high
volume servers, storage, switches or security devices like
Firewalls, Load Balancers and IDS. The SDN controller is
agnostic to the virtualized network elements and can manage or
control the nodes transparently.
1.2 Concepts of Security in SDN and Defence-
in-Depth
With SDN and virtualization of the network, there are various
elements of the network that requires protection. This includes
providing protection to the controller, protecting the physical and
virtual nodes in the network and protecting the inter VM traffic on
the same host from malicious flows. Security of the system also
requires that the controller allows only access to valid users
whose identity is verified and allows trusted application to gain
access to the controller.
Copyright 2015 Page 2 of 7
Also a network design usually has several network segments. It is
important to protect the various network segments from both
internal and external attacks.
An external attack is one that is orchestrated from a source outside
the network. The external attacker may try to attack a network by
various means like inducing malware or take the network down
using a DDoS mechanism. A network can be protected against
external attacks using Firewalls and/or Proxy servers that can
protect the internal network by restricting access for potentially
dangerous external IP addresses and also forcing all traffic to and
from the external world to pass through the proxy server. This
masks the internal nodes from getting exposed to the outside
world and helps protect them from being targets of any DDoS or
DoS attack.
However, there is possibility that attacks may take place from
unknown sources. If there is no previous knowledge of the source
of the malicious attacker, a Firewall cannot restrict access to the
same. In such a situation, the intrusion detection and prevention
devices prove useful. The intrusion detection devices are designed
to detect attacks based on signatures of the packets and thus
complement the other security devices used to secure the network.
These devices may be placed at the gateway to the network to
detect attacks at the network perimeter.
So Firewall, Proxy Servers, Intrusion Detection and Prevention
Services together secure the perimeter of the network.
Apart from attacks originating from external networks, there is
also a possibility of attacks originating from nodes inside the
network. Hence, we need to secure our internal network as well.
This may be done by using early signs of detection of anomalous
behavior in the network or use signature based detection of
attacks in the network. Intrusion detection devices may also be
placed at the entry point of different network segments within the
network for protection from internal attacks. This adds multiple
layers of security and is known as the concept of defence-in-depth
to secure the network from all possible sources.
2. THE PROBLEM STATEMENT AND
IMPLEMENTATION PERSPECTIVE
As outlined in the previous section, the SDN controller is the new
nerve center of the network. The controller abstracts the network
and manages the network from a centralized node. It offers
various networking service, orchestration and management
functions to control the network.
With the controller offering centralized control, it’s pertinent to
utilize the controller as a central monitoring function to manage
the security of the network as well. This provides a definite
opportunity to correlate the network security events at the
controller and managing the network flows to protect the network
from malicious traffic. Since the controller has access to the entire
network, hence, it can influence the protection of the network
from both internal as well as external attacks.
The paper explores the perspective of leveraging this concept to
provide a comprehensive security to the network from all sources
(external or internal to the network). The implementation may be
further enhanced to add security analytics to collect, analyze and
correlate the information from the network, making it operational
as a natural substitute of a SIEM tool. The implementation also
mimics a SOC by taking automating the steps to protect the
network based on the alert based events generated during the
attacks.
3. IDS/IPS SECURITY IN NATIVE
NETWORKS
In this section, we briefly discuss how an Intrusion Detection and
Prevention operates in native networks. This is a precursor to the
security implementation approach that is discussed in later
sections.
An Intrusion Prevention System operates in non-promiscuous
mode. This implies that the device is placed in-line in the network
to listen to packets entering or leaving the network. An IPS works
on the basis of signature based attack detection. In the event of
attack identification, it prevents the packet from passing through
it, thereby protecting the network.
Conversely, an IDS works in promiscuous mode where it listens
to packets mirrored from the network. Based on the rules
configuration, an IDS validates the packets against the signatures
of the packets to identify any attack. In case an attack is identified,
the IDS raises an alert which is fed as input to a SIEM tool. A
SOC is setup where security professionals go through the alerts in
real-time, correlate the information with the help of SIEM tool
and take the required steps towards threat mitigation.
The security professionals also keep a close look out for any
topology changes in the network to identify new nodes which
could be at risk or identify unprotected network segments.
The advantage of IPS is that it can prevent single packet attacks.
If IPS finds a match between the packet contents and the IPS
signatures configured, it is then successfully able to ward off the
attack on the network. However, since IPS has to scan each
packet, match its contents with the rule based signatures and
trigger an alert, therefore, IPS processing impacts the throughput
of the network. Also, if an IPS goes down, it immediately impacts
the network connectivity.
On the other hand, an IDS protects the network re-actively. Once
IDS alert events are generated, the events are correlated at the
SOC and the threat mitigation is initiated. This helps protect
against the action on false-positives thrown up by the IDS as the
alerts are vetted by the security professionals. Also, there is no
impact on the network traffic if an IDS goes down for any reason
since it operates in promiscuous mode.
4. SEPARATION OF SECURITY
PROCESSING FROM PHYSICAL SWITCH
In native networks, there are several IPS/IDS solutions available
that are integrated with the OEM switches for protecting the
network. An example is the Cisco ASA switches integrated with
FirePOWER services (from Sourcefire) and is referred to as Next
Generation IPS that protects the network from various
vulnerabilities. However, with the advent of SDN and NFV
architecture, there is a definite intent to disassociate the data
forwarding plane from the control plane and other security
services provided by the physical infrastructure. This
Copyright 2015 Page 3 of 7
disassociation helps maintain the concept of operational and cost
flexibility that forms the core of SDN-NFV solutions.
5. THE IDPS MONITOR AND
CONTROLLER FOR SDN
As outlined in Sec-3 above, there are definite advantages and
disadvantages of IPS and IDS. The white paper discusses a
solution implementation with IDPS monitor and controller
application that aims to derive the advantages of both worlds as
much as possible. An IDS listens to the packets mirrored from the
network. An application (henceforth referred to as ‘App’)
proactively protects the network from attacks by monitoring the
alerts generated by the IDS in real time. It controls the network by
creating flows to isolate the network from the node that is
identified as the attacker.
Additionally, the solution also aims to protect against flooding
attacks initiated within the network that is not protected by the
IDS. This is achieved by random sampling of the packets flowing
through the network and reporting of the flow and counter
information to the App. This utilizes the sFlow technology that
samples packets on the routers and switches. The App correlates
the information to make intelligent judgments regarding the traffic
flow to detect the flooding attacks. Attacks once identified are
mitigated by blocking the flows originating from the attacker.
Based on the network analysis, the App also aims to tune the IDS
continuously to prevent generation of False positives and keep the
configuration up to date.
Since the App doubles up as a monitor and controller of the
network and aids in threat mitigation, therefore, the App
automates the processes carried out by a SOC. This opens up the
possibility of providing cost advantages by either replacing the
SOC or reducing the resources required to monitor a SOC.
6. IMPLEMENTATION APPROACH
For our concept implementation, the Open Daylight controller is
used along with mininet topology to simulate a network scenario.
Once we bring up the mininet switches and hosts and attach the
mininet switches to the ODL, we configure an interface on the
gateway switch as the ‘Span’ port that mirrors the packets passing
through that switch.
For the purpose of IDS, we install SNORT. Before SNORT IDS is
started, we have to configure the snort.conf file for the host
network that requires to be protected. Additionally, if there are
specific WEB servers or SMTP servers etc, their IP address needs
to be populated as well so that they can be specifically identified
during alert generation. The RULE_PATH along with the set of
rules that requires to be enabled needs to be identified. We also
need to select the output mode for the alert. For our purpose, we
have used text based log and alert generation.
To implement our solution, we have designed an App that resides
within the controller as an OSGi bundle. The App monitors the
alerts from the IDS. The design is flexible to add APIs that are
exposed for any external application or web based request can be
sent to re-program the bundle as required.
In addition to the App, we also have an IDS agent whose role is to
read the details of the snort configuration and send details of the
same to the App as and when requested. Additionally, if there is
any request from the App to reconfigure the IDS, the agent will
have the rights to make the requisite modifications.
For flow sampling, sFlow is configured in the OVS. When traffic
flows, samples are generated from the OVS and the App acts the
collector which receives the samples.
Figure-1 below gives a good overview of the concept
implementation.
Figure 1: Solution Representation
The diagram highlights the IDPS Monitor & Controller residing
inside the ODL controller as an OSGi bundle. The App polls the
text file in which the alerts generated by the IDS are written to in
real-time. The sFlow packet samples from the switch are sent to
the App which processes the information further.
6.1 SNORT Execution and Alert Generation
Once we have the setup in place, we create alerts by carrying out
certain deliberate attacks on the system.
One common form of reconnaissance as a prelude to an attack is
port scanning on the hosts in the system. We can use the nmap
tool to do port scanning of a range of ports in a host system to
identify open ports and the services running on them. This helps
in exposing the possible weaknesses of the system and offer
information platform to launch an attack.
An example of one such nmap scan is shown in Figure-2 below
where a host is scanning for open ports in the 10.0.0.1 machine.
Copyright 2015 Page 4 of 7
Figure 2: Nmap port scan of a network node
The results of the scan show there are no open ports in the host
10.0.0.1 but the very fact that a scan has happened should be good
enough to trigger an alert from the IDS that there is a possible
source address which is trying to do a reconnaissance activity.
If the ‘scan.rules’ is enabled in the configuration file and the right
IP address of HOME_NET address is populated, then SNORT
should throw up an alert as in Figure-3 below:
Figure 3: SNORT Alert
As per the alert generated, it is clear that SNORT has been able to
identify a nmap port scan trying to attempt an Information leak
from the host. The alert clearly identifies the source of the
reconnaissance and the victim machine. Based on the
configuration, the alert is identified as a ‘Priority 2’ alert.
The alert notification is parsed by the App and based on the alert
priority, it is used to determine the threat mitigation action to be
taken.
The App gives an option to configure the blocking period
corresponding to different priority alerts (with higher blocking
period for higher priority alerts) for traffic emanating from the
source address that is found to be generating the attack. Since
addresses can be spoofed, hence it makes sense not to block a
particular IP permanently. However, if the same source IP is
found to be present in the alerts more than ‘n’ times (where ‘n’ is
configurable), then the fraudulent IP is put in the blacklist. An
address once placed in the blacklist would mean that all flows
initiated from the fraudulent source would be permanently
blocked till there is separate administrative action to remove it
from the blacklist.
Figure-4 below shows a flow that is blocked for a period of 120
seconds corresponding to a Priority 1 alert. Post the timeout
period, the flow will be automatically removed and normal traffic
flow can resume.
Figure 4: New Flow based on SNORT Alert
The above example was an illustration of how alerts are generated
and corresponding action is taken by the App in response to the
alerts generated. Similar alerts will be generated when any other
attack is detected and the alert file generation will be similar to the
one shown above.
6.2 Maintaining a White List / Blacklist of IP
Address (es)
In any network, port scanning activity or sending ICMP echo
messages are part of the regular management messages in the
network. However, attacks may actually be disguised in garb of
such regular management messages. In this implementation, the
App takes it upon itself to distinguish between genuine
management messages from the fraudulent activities initiated by
attackers in the network. To this end, its necessary that the App
maintains a White List of IP addresses which are recognized as
genuine IP addresses and activities matching the alert
notifications from such IP addresses are not classified as harmful.
The White List of IP addresses is populated in a file that it read by
the App when it starts up.
The App also exposes an API that can be used to dynamically
add/modify/remove IP addresses from the White List as required.
Just as a White List, we also have a Black list of IP addresses that
are identified to be fraudulent IPs. This Blacklisted IP addresses
are read by the App during initialization. If there are any
addition/modification/deletion to the list at run time, this can be
communicated via an API exposed from the App to do the re-
configuration.
The App, therefore, makes sure that the controller blocks any
flows emanating from such fraudulent, black-listed IP address
upfront.
6.3 Detecting Topology Changes in the
Network
If a new network segment is created, then that should be
automatically detected and probed whether the newly added
network is protected by IDS or not.
The solution has an agent associated with IDS that can share the
Host IP subnets that are monitored and the list of IDS rules
enabled.
The App keeps monitoring the network topology for any additions
/ modifications to the network via the topology manager and host
tracker services in ODL.
If any of the subnets added to the network under the purview of
the controller is not protected, then intimation is sent to intimate
the administrator for corrective measures.
The network Admin can spawn a new VM and run IDS that can
scan the new network segment for possible threats. The alerts
Copyright 2015 Page 5 of 7
generated from the new IDS needs to be sent to the App for future
threat mitigation.
6.4 Dynamic Tuning of the IDS -
Add/Mod/Del IDS Rules
IDS tuning is an absolute necessity for minimizing generation of
False positives. The tuning is a tedious task since it has to be
tuned according to the network configuration. The tuning of the
IDS is a continuous activity if the network configuration changes
dynamically over a period of time.
The App takes the help of the IDS agent to check whether the
snort.conf has configured for the Host Network ($HOME_NET)
to ‘any’. The network subnet to be protected against needs to be
assigned to the Host Network variable to generate definitive
information as per requirement.
The App interacts with an IDS agent to create new customized
rules, as required by the network. The agent interacts with the IDS
to gather information and provides the same back to the App
controlling the threat mitigation. The new rule addition may be a
requirement apart from the rules that come pre-configured with
Snort.
The IDS agent is also instrumental in enabling or disabling rules
in the snort configuration as per the network settings. The
configuration file, may need to dynamically enable or disable
rules based on the traffic pattern in the network. If correct set of
rules are not enabled, the attack detection will not happen.
Conversely, if unnecessary rules are added, it increases the
processing time of the IDS. Hence, rule optimization is necessary
for efficient processing of the IDS.
Additionally, the App applies a hook on the controller to check
for the type of traffic packets exchanged in the network. If any
traffic is generated which does not conform to the protocol types
allowed, then the App introduces flows to block traffic from the
source matching the protocol - so that the traffic flow with that
protocol is blocked immediately. If the protocol is to be later
allowed, then the flow matching the protocol type needs to be
removed by the App.
Conversely, we may have rules pre-configured to prevent traffic
for a particular protocol. But if any such traffic is generated as per
genuine network requirement, then the source IP address for such
traffic should be placed in the ‘White List’ to notify the App that
such type of traffic should be allowed and legal. The App will
then detect such traffic and will desist from introducing a
blocking flow for the source IP even though it matches a blocked
protocol. Hence, the hosts whose IP address is included in the
white listed IP address, have the liberty to carry any activity
which may be blocked for other hosts.
6.5 sFlow Based Detection of Flood Attacks in
Internal Network
We have briefly discussed the sFlow integration in earlier
sections. In this section, we give a detailed outline of sFlow
integration with our application and the logic around it.
sFlow is sampled flow [3]. It used for random sampling of packets
in high speed networks. sFlow has two components - one is the
sFlow agent and the other is the sFlow collector. There are several
device manufacturers that support sFlow which also includes the
Open VSwitch (which is used in mininet setup). The sFlow agent
is inbuilt in the device and needs to be configured when bringing
up the switch. However, there needs to be a separate ‘collector’
which listens to the sFlow sample packets and utilizes the sample
flow to make some intelligent judgment regarding the traffic flow
in the network.
In the concept implementation, we have configured for sFlow in
the switches which cater to traffic internal to the network. This is
to track the traffic that does not flow through the gateway switch.
Hence, any violations by traffic packets in the internal network
will not be detected by the security devices at the perimeter. This
leaves the internal network open to attacks. It’s precisely for this
reason that we configure sFlow and monitor the network.
A flow in the sFlow context is defined as traffic from the Data
Source to the Data destination. For example, traffic flow from
Data Source - say IP_1 to Data Destination - say IP_2, is treated
as one flow while traffic from another Data Source - say IP_3 to
IP_2, is treated as a separate flow. A flow sample contains the
details of the Layer-2 to Layer-4 details of the packet exchanged
along with the size of the packet that is exchanged.
In our solution, the App which is running as an OSGi bundle
within the controller, acts as the collector of the sFlow packets. It
listens to the port ‘::6343’ for packets and their information
sampled by sFlow is sent to the collector.
The sFlow sampling logic is based on the concept of probability
based sampling. If there is large number of packets sampled, the
packets from different flows in the network gives a deterministic
representation of the actual packet count corresponding to each
flow.
If there are a total number of packets, N, and a total number of
samples, n, of which a certain number belong to a particular flow,
f, then the number of frames in the flow, Nf, can be estimated
using the equation: Nf = (f/n) * N [3].
According to statistical calculations, the percentage error
corresponding to a flow is depicted as:
%error ≤ K * √(1/f) (where K is a constant) [3]
Hence, as ‘f’ increases, the error possibility corresponding to a
flow goes down exponentially.
While configuring for sFlow, there needs to be sampling rate and
polling frequency mentioned. The sampling rate determines how
frequently the flows are sampled. For example, if the sampling
rate is configured as 10, then it implies that one out of 10 flow
packets will be sampled and sent to the collector. Based on the
traffic rate flowing through a switch, the sFlow sampling needs to
be configured accordingly. This should ensure that sufficiently
large number of packets is sampled and calculations based on
such flows mirrors the actual packet flows in the switch.
Figure-5 below shows a flow sample collected and its contents in
the Raw packet header:
Copyright 2015 Page 6 of 7
Figure 5: Raw packet header from Flow sample
On the other hand, the counter samples are used to obtain the
number of input and output packets that have been exchanged on
an interface over a period of time.
The Figure-6 below shows a counter sample collection:
Figure 6: Counter Sample contents
The sFlow configuration on the switch needs to ensure that a large
number of samples is collected so that it is reflective of the actual
traffic flow in the network. The logic applied in the App
(collector) is to average the data packet sample information over a
period of time (with continuously changing reference window) so
that we are able to deduce the average bandwidth utilization over
different interfaces.
The bandwidth utilization in each interface is calculated and
compared against the threshold bandwidth utilization levels for
each interface. If the bandwidth utilization by a flow is found to
be exceeding the threshold value, then the source address of the
flow is blocked. Exceeding the threshold bandwidth is assumed to
be a case of flooding the destination host interface. The period of
blocking of the interface is made configurable for the network
admin to decide. Also, the period of calculation of the average
bandwidth utilization on an interface can be configured by the
administrator user.
The App also takes care to calculate the average bandwidth
utilization levels over a period of time to ensure that sudden
spikes in usage even out and does not cause unnecessary blockage
of the source.
7. SOLUTION DIFFERENTIATORS
Having outlined the implementation strategy, its important to
highlight the solution differentiators and the Capex and Opex
benefits associated with it.
7.1 Pro-active Mitigation with Zero Impact on
Traffic
One of the major drawbacks of deploying an in-band security
appliance in the network is that it affects the traffic throughput of
the interface to which it is attached, thereby becoming a
bottleneck for the network that is protected. Conversely, there is
an obvious advantage as well since in-band devices (usually
placed at the gateway of a network) can match the packet
signature of the incoming packets on the wire before it can intrude
the network. Security devices deployed in-band and having threat
mitigation capability (eg. IPS), offer a pro-active mechanism of
protecting the network whereas out-of-band devices (eg. IDS)
offer re-active security.
In the IDPS solution implementation, the SDN controller controls
the entire network. Hence, the intrusion prevention role is played
out by the controller although threat detection is made by IDS
which operates in promiscuous mode. So not only is our security
mechanism robust in terms of providing pro-active mitigation but
additionally provides non-intrusive threat detection capability as
well. This is the uniqueness of the solution that is sought to be
highlighted in this paper.
7.2 Low cost offering for ODL (CAPEX
Saving)
With software based IDS, no proprietary hardware and
automation of security configuration, the implemented solution
provides a low cost offering with ease of Deployment in SDN.
Presently there is only an offering by Radware which is named
‘Defense4all’ to secure SDN networks[4]. However, the solution
requires use of a proprietary “Attack Mitigation System” which
does the scrubbing of the packets that are identified to be possible
perpetrators of an attack. Additionally, it needs to be mentioned
that ‘Defense4all’ is a DoS mitigation solution only.
The IDPS application, in conjunction with IDS, offers protection
against reconnaissance attempts, malware detection, DDoS attacks
and many more. Also the system offers ease of deployment with
automation of the IDS tuning based on network traffic patterns.
Logs and alert information thrown by IDS coupled with statistical
information collection at the controller provides for actionable
security intelligence to reduce time for collating and correlating
threat information. With automated threat analysis, this low cost
solution is a replacement for Security Information and Event
Management tool, thereby providing huge potential for CAPEX
savings.
Copyright 2015 Page 7 of 7
7.3 Ease of Deployment (OPEX Saving)
The IDPS application is developed as a plugin that can be easily
installed within the ODL framework as an OSGi bundle. The
application is agnostic to the security device used in the network.
It can be integrated with any security appliance with
customization of the module that reads the alert file for processing
of the threats from the network. Additionally, the automated IDS
tuning capability provided by the IDS agent also aids in the
deployment process of the IDS that helps to keep it in sync with
the network traffic pattern and weeds out unnecessary processing
of signatures that are not required. This helps in speeding up the
processing of the signature based attack detection at the IDS.
With an obvious ease of deployment, actionable intelligence and
most importantly, proactive threat mitigation, the solution helps in
replicating a security operations center which brings in the
opportunity of huge OPEX savings.
8. CONCLUSION
The implementation approach tries to encompass the security in
SDN based networks. We have already discussed in depth about
the rationale of using an IDS vis-à-vis an IPS for building our
solution and the benefits to be reaped out of it.
However, it must be mentioned that the system still lacks the
defense against single packet attacks that will have already
impacted the network before its detected. These attacks might be
protected against by using a Host IDS or any end-point security
that is deemed suitable. Also, there is a bound to be a time lag
between generation of the alerts and their processing since the
alerts need to be polled by the application before they are
processed. However, this time lag is not significant to be a cause
of concern since the polling of the alert information is frequent
and ensures that the packet is picked up for processing almost
immediately as the alert is generated. Also, it must be kept in
mind that reasonably well tuned IDS will not throw up an awful
lot of alerts. Hence, the system is not much worse off in terms of
attack response but significantly better off in terms of removing
the network bottleneck that is brought about by an IPS.
With the future of networking bending towards SDN and
virtualization, it’s important to have a multi-pronged approach to
solve the challenges thrown by the new architecture. This
implementation has been an attempt towards that end to combine
the features of traditional IDS with an application to convert a re-
active design to a proactive one through automation and also
bring in the flavor of sampling based analytics to aid in protecting
the network.
The solution is by no means touted as a comprehensive solution
for SDN networks. As already mentioned earlier, securing SDN
networks has a wide range of challenges which include (but not
limiting to) securing of the North and South Bound interfaces of
the controller, securing the controller itself from external attacks,
securing the VNFs, establishing trust between the elements of the
SDN network and many more. The paper wishes to draw the
attention of the reader towards a prospective seeded solution that
could be handy towards building a holistic security design for
SDN networks. The proposed solution is intended to be the initial
spade work towards a larger goal of a fool-proof security of the
entire SDN eco-system.
9. ABBREVIATIONS
API – Application Programming Interface
DDoS – Distributed Denial of Service
DoS – Denial of Service
IDS - Intrusion Detection System
IDPS – Intrusion Detection and Prevention System
IPS - Intrusion Prevention System
NFV - Network Function Virtualisation
OVS – Open VSwtch
ODL - Open Day Light
OSGi - Open Service Gateway initiative
SDN - Software Defined Networking
SIEM - Security Information and Event Management
SOC - Security Operations Center
VM - Virtual Machine
10. REFERENCES
[1] http://blogs.gartner.com/andrew-
lerner/2014/12/08/predicting-sdn-adoption/.
[2] http://newsroom.juniper.net/press-releases/new-juniper-
networks-study-finds-u-s-companies-sp-nyse-jnpr-1134411
[3] http://www.sFlow.org/sFlow_version_5.txt.
http://www.sFlow.org/packetSamplingBasics
[4] https://wiki.opendaylight.org/view/Defense4All:Tutorial

Mais conteúdo relacionado

Mais procurados

IRJET - IDS for Wifi Security
IRJET -  	  IDS for Wifi SecurityIRJET -  	  IDS for Wifi Security
IRJET - IDS for Wifi SecurityIRJET Journal
 
REAL-TIME INTRUSION DETECTION SYSTEM FOR BIG DATA
REAL-TIME INTRUSION DETECTION SYSTEM FOR BIG DATAREAL-TIME INTRUSION DETECTION SYSTEM FOR BIG DATA
REAL-TIME INTRUSION DETECTION SYSTEM FOR BIG DATAijp2p
 
Intrusion preventionintrusion detection
Intrusion preventionintrusion detectionIntrusion preventionintrusion detection
Intrusion preventionintrusion detectionIJCNCJournal
 
Intrusion Detection and Prevention System in an Enterprise Network
Intrusion Detection and Prevention System in an Enterprise NetworkIntrusion Detection and Prevention System in an Enterprise Network
Intrusion Detection and Prevention System in an Enterprise NetworkOkehie Collins
 
Intrusion detection and prevention system for network using Honey pots and Ho...
Intrusion detection and prevention system for network using Honey pots and Ho...Intrusion detection and prevention system for network using Honey pots and Ho...
Intrusion detection and prevention system for network using Honey pots and Ho...Eng. Mohammed Ahmed Siddiqui
 
CRASHOVERRIDE Analysis of the Threat to Electric Grid Operations. Cyber-attac...
CRASHOVERRIDE Analysis of the Threat to Electric Grid Operations. Cyber-attac...CRASHOVERRIDE Analysis of the Threat to Electric Grid Operations. Cyber-attac...
CRASHOVERRIDE Analysis of the Threat to Electric Grid Operations. Cyber-attac...Muhammad FAHAD
 
ClubHack Magazine issue 26 March 2012
ClubHack Magazine issue 26 March 2012ClubHack Magazine issue 26 March 2012
ClubHack Magazine issue 26 March 2012ClubHack
 
Detection of Rogue Access Point in WLAN using Hopfield Neural Network
Detection of Rogue Access Point in WLAN using Hopfield Neural Network  Detection of Rogue Access Point in WLAN using Hopfield Neural Network
Detection of Rogue Access Point in WLAN using Hopfield Neural Network IJECEIAES
 
A secure intrusion detection system against ddos attack in wireless mobile ad...
A secure intrusion detection system against ddos attack in wireless mobile ad...A secure intrusion detection system against ddos attack in wireless mobile ad...
A secure intrusion detection system against ddos attack in wireless mobile ad...vishnuRajan20
 
Nice network intrusion detection and countermeasure
Nice network intrusion detection and countermeasureNice network intrusion detection and countermeasure
Nice network intrusion detection and countermeasureIEEEFINALYEARPROJECTS
 
Intrusion detection system
Intrusion detection system Intrusion detection system
Intrusion detection system gaurav koriya
 
An Extensive Survey of Intrusion Detection Systems
An Extensive Survey of Intrusion Detection SystemsAn Extensive Survey of Intrusion Detection Systems
An Extensive Survey of Intrusion Detection SystemsIRJET Journal
 
Survey on Host and Network Based Intrusion Detection System
Survey on Host and Network Based Intrusion Detection SystemSurvey on Host and Network Based Intrusion Detection System
Survey on Host and Network Based Intrusion Detection SystemEswar Publications
 
Physical/Network Access Control
Physical/Network Access ControlPhysical/Network Access Control
Physical/Network Access Controljwpiccininni
 
D03302030036
D03302030036D03302030036
D03302030036theijes
 
RSAC 2021 Spelunking Through the Steps of a Control System Hack
RSAC 2021 Spelunking Through the Steps of a Control System HackRSAC 2021 Spelunking Through the Steps of a Control System Hack
RSAC 2021 Spelunking Through the Steps of a Control System HackDan Gunter
 

Mais procurados (19)

IRJET - IDS for Wifi Security
IRJET -  	  IDS for Wifi SecurityIRJET -  	  IDS for Wifi Security
IRJET - IDS for Wifi Security
 
REAL-TIME INTRUSION DETECTION SYSTEM FOR BIG DATA
REAL-TIME INTRUSION DETECTION SYSTEM FOR BIG DATAREAL-TIME INTRUSION DETECTION SYSTEM FOR BIG DATA
REAL-TIME INTRUSION DETECTION SYSTEM FOR BIG DATA
 
Intrusion preventionintrusion detection
Intrusion preventionintrusion detectionIntrusion preventionintrusion detection
Intrusion preventionintrusion detection
 
Intrusion Detection and Prevention System in an Enterprise Network
Intrusion Detection and Prevention System in an Enterprise NetworkIntrusion Detection and Prevention System in an Enterprise Network
Intrusion Detection and Prevention System in an Enterprise Network
 
Intrusion detection and prevention system for network using Honey pots and Ho...
Intrusion detection and prevention system for network using Honey pots and Ho...Intrusion detection and prevention system for network using Honey pots and Ho...
Intrusion detection and prevention system for network using Honey pots and Ho...
 
CRASHOVERRIDE Analysis of the Threat to Electric Grid Operations. Cyber-attac...
CRASHOVERRIDE Analysis of the Threat to Electric Grid Operations. Cyber-attac...CRASHOVERRIDE Analysis of the Threat to Electric Grid Operations. Cyber-attac...
CRASHOVERRIDE Analysis of the Threat to Electric Grid Operations. Cyber-attac...
 
Mobile slide
Mobile slideMobile slide
Mobile slide
 
ClubHack Magazine issue 26 March 2012
ClubHack Magazine issue 26 March 2012ClubHack Magazine issue 26 March 2012
ClubHack Magazine issue 26 March 2012
 
Review of network diagram
Review of network diagramReview of network diagram
Review of network diagram
 
Detection of Rogue Access Point in WLAN using Hopfield Neural Network
Detection of Rogue Access Point in WLAN using Hopfield Neural Network  Detection of Rogue Access Point in WLAN using Hopfield Neural Network
Detection of Rogue Access Point in WLAN using Hopfield Neural Network
 
06686259 20140405 205404
06686259 20140405 20540406686259 20140405 205404
06686259 20140405 205404
 
A secure intrusion detection system against ddos attack in wireless mobile ad...
A secure intrusion detection system against ddos attack in wireless mobile ad...A secure intrusion detection system against ddos attack in wireless mobile ad...
A secure intrusion detection system against ddos attack in wireless mobile ad...
 
Nice network intrusion detection and countermeasure
Nice network intrusion detection and countermeasureNice network intrusion detection and countermeasure
Nice network intrusion detection and countermeasure
 
Intrusion detection system
Intrusion detection system Intrusion detection system
Intrusion detection system
 
An Extensive Survey of Intrusion Detection Systems
An Extensive Survey of Intrusion Detection SystemsAn Extensive Survey of Intrusion Detection Systems
An Extensive Survey of Intrusion Detection Systems
 
Survey on Host and Network Based Intrusion Detection System
Survey on Host and Network Based Intrusion Detection SystemSurvey on Host and Network Based Intrusion Detection System
Survey on Host and Network Based Intrusion Detection System
 
Physical/Network Access Control
Physical/Network Access ControlPhysical/Network Access Control
Physical/Network Access Control
 
D03302030036
D03302030036D03302030036
D03302030036
 
RSAC 2021 Spelunking Through the Steps of a Control System Hack
RSAC 2021 Spelunking Through the Steps of a Control System HackRSAC 2021 Spelunking Through the Steps of a Control System Hack
RSAC 2021 Spelunking Through the Steps of a Control System Hack
 

Semelhante a TACTiCS_WP Security_Addressing Security in SDN Environment

network_security.docx_2.pdf
network_security.docx_2.pdfnetwork_security.docx_2.pdf
network_security.docx_2.pdfahmed53254
 
Network Security Fundamentals
Network Security FundamentalsNetwork Security Fundamentals
Network Security FundamentalsRahmat Suhatman
 
Light sec for service providers brochure
Light sec for service providers brochureLight sec for service providers brochure
Light sec for service providers brochureGeorge Wainblat
 
Gigamon - Network Visibility Solutions
Gigamon - Network Visibility SolutionsGigamon - Network Visibility Solutions
Gigamon - Network Visibility SolutionsTom Kopko
 
DEPLOYMENT OF INTRUSION PREVENTION SYSTEM ON MULTI-CORE PROCESSOR BASED SECUR...
DEPLOYMENT OF INTRUSION PREVENTION SYSTEM ON MULTI-CORE PROCESSOR BASED SECUR...DEPLOYMENT OF INTRUSION PREVENTION SYSTEM ON MULTI-CORE PROCESSOR BASED SECUR...
DEPLOYMENT OF INTRUSION PREVENTION SYSTEM ON MULTI-CORE PROCESSOR BASED SECUR...IJCNCJournal
 
Optimized Intrusion Detection System using Deep Learning Algorithm
Optimized Intrusion Detection System using Deep Learning AlgorithmOptimized Intrusion Detection System using Deep Learning Algorithm
Optimized Intrusion Detection System using Deep Learning Algorithmijtsrd
 
Whitepaper - Software Defined Networking for the Telco Industry
Whitepaper - Software Defined Networking for the Telco IndustryWhitepaper - Software Defined Networking for the Telco Industry
Whitepaper - Software Defined Networking for the Telco Industryaap3 IT Recruitment
 
AN ISP BASED NOTIFICATION AND DETECTION SYSTEM TO MAXIMIZE EFFICIENCY OF CLIE...
AN ISP BASED NOTIFICATION AND DETECTION SYSTEM TO MAXIMIZE EFFICIENCY OF CLIE...AN ISP BASED NOTIFICATION AND DETECTION SYSTEM TO MAXIMIZE EFFICIENCY OF CLIE...
AN ISP BASED NOTIFICATION AND DETECTION SYSTEM TO MAXIMIZE EFFICIENCY OF CLIE...IJNSA Journal
 
Unified Security Plugin for Opendaylight Controller
Unified Security Plugin for Opendaylight ControllerUnified Security Plugin for Opendaylight Controller
Unified Security Plugin for Opendaylight ControllerSaikat Chaudhuri
 
Secure intrusion detection and countermeasure selection in virtual system usi...
Secure intrusion detection and countermeasure selection in virtual system usi...Secure intrusion detection and countermeasure selection in virtual system usi...
Secure intrusion detection and countermeasure selection in virtual system usi...eSAT Publishing House
 
Enterprise firewalls feature and benefits
Enterprise firewalls feature and benefitsEnterprise firewalls feature and benefits
Enterprise firewalls feature and benefitsAnthony Daniel
 
UNCONSTRAINED ENDPOINT SECURITY SYSTEM: UEPTSS
UNCONSTRAINED ENDPOINT SECURITY SYSTEM: UEPTSSUNCONSTRAINED ENDPOINT SECURITY SYSTEM: UEPTSS
UNCONSTRAINED ENDPOINT SECURITY SYSTEM: UEPTSSIJNSA Journal
 
UNCONSTRAINED ENDPOINT SECURITY SYSTEM: UEPTSS
UNCONSTRAINED ENDPOINT SECURITY SYSTEM: UEPTSSUNCONSTRAINED ENDPOINT SECURITY SYSTEM: UEPTSS
UNCONSTRAINED ENDPOINT SECURITY SYSTEM: UEPTSSIJNSA Journal
 
Toward Continuous Cybersecurity with Network Automation
Toward Continuous Cybersecurity with Network AutomationToward Continuous Cybersecurity with Network Automation
Toward Continuous Cybersecurity with Network AutomationE.S.G. JR. Consulting, Inc.
 
Toward Continuous Cybersecurity With Network Automation
Toward Continuous Cybersecurity With Network AutomationToward Continuous Cybersecurity With Network Automation
Toward Continuous Cybersecurity With Network AutomationKen Flott
 
unit 2 IT security solution.pptx
unit 2 IT security solution.pptxunit 2 IT security solution.pptx
unit 2 IT security solution.pptxlochanrajdahal
 
Designing Security Assessment of Client Server System using Attack Tree Modeling
Designing Security Assessment of Client Server System using Attack Tree ModelingDesigning Security Assessment of Client Server System using Attack Tree Modeling
Designing Security Assessment of Client Server System using Attack Tree Modelingijtsrd
 

Semelhante a TACTiCS_WP Security_Addressing Security in SDN Environment (20)

network_security.docx_2.pdf
network_security.docx_2.pdfnetwork_security.docx_2.pdf
network_security.docx_2.pdf
 
Network Security Fundamentals
Network Security FundamentalsNetwork Security Fundamentals
Network Security Fundamentals
 
Light sec for service providers brochure
Light sec for service providers brochureLight sec for service providers brochure
Light sec for service providers brochure
 
Gigamon - Network Visibility Solutions
Gigamon - Network Visibility SolutionsGigamon - Network Visibility Solutions
Gigamon - Network Visibility Solutions
 
DEPLOYMENT OF INTRUSION PREVENTION SYSTEM ON MULTI-CORE PROCESSOR BASED SECUR...
DEPLOYMENT OF INTRUSION PREVENTION SYSTEM ON MULTI-CORE PROCESSOR BASED SECUR...DEPLOYMENT OF INTRUSION PREVENTION SYSTEM ON MULTI-CORE PROCESSOR BASED SECUR...
DEPLOYMENT OF INTRUSION PREVENTION SYSTEM ON MULTI-CORE PROCESSOR BASED SECUR...
 
Optimized Intrusion Detection System using Deep Learning Algorithm
Optimized Intrusion Detection System using Deep Learning AlgorithmOptimized Intrusion Detection System using Deep Learning Algorithm
Optimized Intrusion Detection System using Deep Learning Algorithm
 
Network srcurity
Network srcurityNetwork srcurity
Network srcurity
 
Whitepaper - Software Defined Networking for the Telco Industry
Whitepaper - Software Defined Networking for the Telco IndustryWhitepaper - Software Defined Networking for the Telco Industry
Whitepaper - Software Defined Networking for the Telco Industry
 
PACE-IT, Security+1.1: Introduction to Network Devices (part 2)
PACE-IT, Security+1.1: Introduction to Network Devices (part 2)PACE-IT, Security+1.1: Introduction to Network Devices (part 2)
PACE-IT, Security+1.1: Introduction to Network Devices (part 2)
 
AN ISP BASED NOTIFICATION AND DETECTION SYSTEM TO MAXIMIZE EFFICIENCY OF CLIE...
AN ISP BASED NOTIFICATION AND DETECTION SYSTEM TO MAXIMIZE EFFICIENCY OF CLIE...AN ISP BASED NOTIFICATION AND DETECTION SYSTEM TO MAXIMIZE EFFICIENCY OF CLIE...
AN ISP BASED NOTIFICATION AND DETECTION SYSTEM TO MAXIMIZE EFFICIENCY OF CLIE...
 
Unified Security Plugin for Opendaylight Controller
Unified Security Plugin for Opendaylight ControllerUnified Security Plugin for Opendaylight Controller
Unified Security Plugin for Opendaylight Controller
 
Secure intrusion detection and countermeasure selection in virtual system usi...
Secure intrusion detection and countermeasure selection in virtual system usi...Secure intrusion detection and countermeasure selection in virtual system usi...
Secure intrusion detection and countermeasure selection in virtual system usi...
 
Enterprise firewalls feature and benefits
Enterprise firewalls feature and benefitsEnterprise firewalls feature and benefits
Enterprise firewalls feature and benefits
 
UNCONSTRAINED ENDPOINT SECURITY SYSTEM: UEPTSS
UNCONSTRAINED ENDPOINT SECURITY SYSTEM: UEPTSSUNCONSTRAINED ENDPOINT SECURITY SYSTEM: UEPTSS
UNCONSTRAINED ENDPOINT SECURITY SYSTEM: UEPTSS
 
UNCONSTRAINED ENDPOINT SECURITY SYSTEM: UEPTSS
UNCONSTRAINED ENDPOINT SECURITY SYSTEM: UEPTSSUNCONSTRAINED ENDPOINT SECURITY SYSTEM: UEPTSS
UNCONSTRAINED ENDPOINT SECURITY SYSTEM: UEPTSS
 
Toward Continuous Cybersecurity with Network Automation
Toward Continuous Cybersecurity with Network AutomationToward Continuous Cybersecurity with Network Automation
Toward Continuous Cybersecurity with Network Automation
 
Toward Continuous Cybersecurity With Network Automation
Toward Continuous Cybersecurity With Network AutomationToward Continuous Cybersecurity With Network Automation
Toward Continuous Cybersecurity With Network Automation
 
unit 2 IT security solution.pptx
unit 2 IT security solution.pptxunit 2 IT security solution.pptx
unit 2 IT security solution.pptx
 
Designing Security Assessment of Client Server System using Attack Tree Modeling
Designing Security Assessment of Client Server System using Attack Tree ModelingDesigning Security Assessment of Client Server System using Attack Tree Modeling
Designing Security Assessment of Client Server System using Attack Tree Modeling
 
Network security
Network securityNetwork security
Network security
 

TACTiCS_WP Security_Addressing Security in SDN Environment

  • 1. Copyright 2015 Page 1 of 7 Addressing Security in SDN Environment: A Perspective Saikat Chaudhuri Tata Consultancy Services, Bangalore +91-8884742237 chaudhuri.saikat@tcs.com Subash Adigal Tata Consultancy Services, Delhi +91-9643330517 subash.adigal@tcs.com ABSTRACT For the last two years, there is a definite buzz around SDN which is complemented by NFV. The strong focus on SDN is vindicated by the fact that by 2016, there is a prediction that more than 10,000 enterprises will deploy SDN in their networks, which is almost a ten-fold increase from the data in 2014 [1]. Also, according to a survey done by Juniper in 2014, out of 400 U.S. Businesses, around 52.5 percent of the companies plan to adopt SDN in the future. 27 percent out of these companies have stated that they are ready or almost ready to adopt SDN [2]. However, along with the future prospect of SDN, there are numerous concerns surrounding the security aspect associated with it. Despite the concerns, organizations are keen to adopt SDN due to its numerous benefits amongst which analytics, reporting and resulting automation are some of the key elements. The White Paper discusses a possible solution towards addressing the security concern by automating the security in SDN environment. The solution deploys an application that continuously monitors the alerts generated by IDS, analyses the reports and automates action to block flows from hosts based on the alerts raised by the IDS. The paper also discusses the implementation of sFlow based monitoring of the network. This is based on packet sampling which is utilized to analyze the traffic trend and determine if there is a possible flooding attack which may have gone undetected in the network. Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice. To copy otherwise, or republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee. TACTiCS – TCS Technical Architects’ Conference 2012 1. INTRODUCTION 1.1 SDN and NFV SDN is a term coined to exhibit the concept in networking where the control plane is decoupled from the physical infrastructure and is instead moved out into a separate entity called the controller. The controller provides networking services that help to manage the network and also determine how the data plane flows are configured. Since the controller runs as a separate software application to manage the entire network, hence its termed as ‘Software Defined Networking’ aka SDN. The switches (physical infrastructure), that handle the traffic flow, may be made available from multiple vendors as long as they are compatible with the protocol of communication between the controller and the switch. This reduces dependency on any particular equipment manufacturer and the service provider has the flexibility to use switches of their choice that gives them the cost and/or operational advantage. NFV complements the concept of SDN by providing the additional flexibility of dynamically providing network functions on VMs. This enhances the levels of utilization of the physical hardware and also bring in cost benefits where number of VMs can be scaled up or down dynamically as per the requirement. The VMs may be spawned on top of physical hosts such as high volume servers, storage, switches or security devices like Firewalls, Load Balancers and IDS. The SDN controller is agnostic to the virtualized network elements and can manage or control the nodes transparently. 1.2 Concepts of Security in SDN and Defence- in-Depth With SDN and virtualization of the network, there are various elements of the network that requires protection. This includes providing protection to the controller, protecting the physical and virtual nodes in the network and protecting the inter VM traffic on the same host from malicious flows. Security of the system also requires that the controller allows only access to valid users whose identity is verified and allows trusted application to gain access to the controller.
  • 2. Copyright 2015 Page 2 of 7 Also a network design usually has several network segments. It is important to protect the various network segments from both internal and external attacks. An external attack is one that is orchestrated from a source outside the network. The external attacker may try to attack a network by various means like inducing malware or take the network down using a DDoS mechanism. A network can be protected against external attacks using Firewalls and/or Proxy servers that can protect the internal network by restricting access for potentially dangerous external IP addresses and also forcing all traffic to and from the external world to pass through the proxy server. This masks the internal nodes from getting exposed to the outside world and helps protect them from being targets of any DDoS or DoS attack. However, there is possibility that attacks may take place from unknown sources. If there is no previous knowledge of the source of the malicious attacker, a Firewall cannot restrict access to the same. In such a situation, the intrusion detection and prevention devices prove useful. The intrusion detection devices are designed to detect attacks based on signatures of the packets and thus complement the other security devices used to secure the network. These devices may be placed at the gateway to the network to detect attacks at the network perimeter. So Firewall, Proxy Servers, Intrusion Detection and Prevention Services together secure the perimeter of the network. Apart from attacks originating from external networks, there is also a possibility of attacks originating from nodes inside the network. Hence, we need to secure our internal network as well. This may be done by using early signs of detection of anomalous behavior in the network or use signature based detection of attacks in the network. Intrusion detection devices may also be placed at the entry point of different network segments within the network for protection from internal attacks. This adds multiple layers of security and is known as the concept of defence-in-depth to secure the network from all possible sources. 2. THE PROBLEM STATEMENT AND IMPLEMENTATION PERSPECTIVE As outlined in the previous section, the SDN controller is the new nerve center of the network. The controller abstracts the network and manages the network from a centralized node. It offers various networking service, orchestration and management functions to control the network. With the controller offering centralized control, it’s pertinent to utilize the controller as a central monitoring function to manage the security of the network as well. This provides a definite opportunity to correlate the network security events at the controller and managing the network flows to protect the network from malicious traffic. Since the controller has access to the entire network, hence, it can influence the protection of the network from both internal as well as external attacks. The paper explores the perspective of leveraging this concept to provide a comprehensive security to the network from all sources (external or internal to the network). The implementation may be further enhanced to add security analytics to collect, analyze and correlate the information from the network, making it operational as a natural substitute of a SIEM tool. The implementation also mimics a SOC by taking automating the steps to protect the network based on the alert based events generated during the attacks. 3. IDS/IPS SECURITY IN NATIVE NETWORKS In this section, we briefly discuss how an Intrusion Detection and Prevention operates in native networks. This is a precursor to the security implementation approach that is discussed in later sections. An Intrusion Prevention System operates in non-promiscuous mode. This implies that the device is placed in-line in the network to listen to packets entering or leaving the network. An IPS works on the basis of signature based attack detection. In the event of attack identification, it prevents the packet from passing through it, thereby protecting the network. Conversely, an IDS works in promiscuous mode where it listens to packets mirrored from the network. Based on the rules configuration, an IDS validates the packets against the signatures of the packets to identify any attack. In case an attack is identified, the IDS raises an alert which is fed as input to a SIEM tool. A SOC is setup where security professionals go through the alerts in real-time, correlate the information with the help of SIEM tool and take the required steps towards threat mitigation. The security professionals also keep a close look out for any topology changes in the network to identify new nodes which could be at risk or identify unprotected network segments. The advantage of IPS is that it can prevent single packet attacks. If IPS finds a match between the packet contents and the IPS signatures configured, it is then successfully able to ward off the attack on the network. However, since IPS has to scan each packet, match its contents with the rule based signatures and trigger an alert, therefore, IPS processing impacts the throughput of the network. Also, if an IPS goes down, it immediately impacts the network connectivity. On the other hand, an IDS protects the network re-actively. Once IDS alert events are generated, the events are correlated at the SOC and the threat mitigation is initiated. This helps protect against the action on false-positives thrown up by the IDS as the alerts are vetted by the security professionals. Also, there is no impact on the network traffic if an IDS goes down for any reason since it operates in promiscuous mode. 4. SEPARATION OF SECURITY PROCESSING FROM PHYSICAL SWITCH In native networks, there are several IPS/IDS solutions available that are integrated with the OEM switches for protecting the network. An example is the Cisco ASA switches integrated with FirePOWER services (from Sourcefire) and is referred to as Next Generation IPS that protects the network from various vulnerabilities. However, with the advent of SDN and NFV architecture, there is a definite intent to disassociate the data forwarding plane from the control plane and other security services provided by the physical infrastructure. This
  • 3. Copyright 2015 Page 3 of 7 disassociation helps maintain the concept of operational and cost flexibility that forms the core of SDN-NFV solutions. 5. THE IDPS MONITOR AND CONTROLLER FOR SDN As outlined in Sec-3 above, there are definite advantages and disadvantages of IPS and IDS. The white paper discusses a solution implementation with IDPS monitor and controller application that aims to derive the advantages of both worlds as much as possible. An IDS listens to the packets mirrored from the network. An application (henceforth referred to as ‘App’) proactively protects the network from attacks by monitoring the alerts generated by the IDS in real time. It controls the network by creating flows to isolate the network from the node that is identified as the attacker. Additionally, the solution also aims to protect against flooding attacks initiated within the network that is not protected by the IDS. This is achieved by random sampling of the packets flowing through the network and reporting of the flow and counter information to the App. This utilizes the sFlow technology that samples packets on the routers and switches. The App correlates the information to make intelligent judgments regarding the traffic flow to detect the flooding attacks. Attacks once identified are mitigated by blocking the flows originating from the attacker. Based on the network analysis, the App also aims to tune the IDS continuously to prevent generation of False positives and keep the configuration up to date. Since the App doubles up as a monitor and controller of the network and aids in threat mitigation, therefore, the App automates the processes carried out by a SOC. This opens up the possibility of providing cost advantages by either replacing the SOC or reducing the resources required to monitor a SOC. 6. IMPLEMENTATION APPROACH For our concept implementation, the Open Daylight controller is used along with mininet topology to simulate a network scenario. Once we bring up the mininet switches and hosts and attach the mininet switches to the ODL, we configure an interface on the gateway switch as the ‘Span’ port that mirrors the packets passing through that switch. For the purpose of IDS, we install SNORT. Before SNORT IDS is started, we have to configure the snort.conf file for the host network that requires to be protected. Additionally, if there are specific WEB servers or SMTP servers etc, their IP address needs to be populated as well so that they can be specifically identified during alert generation. The RULE_PATH along with the set of rules that requires to be enabled needs to be identified. We also need to select the output mode for the alert. For our purpose, we have used text based log and alert generation. To implement our solution, we have designed an App that resides within the controller as an OSGi bundle. The App monitors the alerts from the IDS. The design is flexible to add APIs that are exposed for any external application or web based request can be sent to re-program the bundle as required. In addition to the App, we also have an IDS agent whose role is to read the details of the snort configuration and send details of the same to the App as and when requested. Additionally, if there is any request from the App to reconfigure the IDS, the agent will have the rights to make the requisite modifications. For flow sampling, sFlow is configured in the OVS. When traffic flows, samples are generated from the OVS and the App acts the collector which receives the samples. Figure-1 below gives a good overview of the concept implementation. Figure 1: Solution Representation The diagram highlights the IDPS Monitor & Controller residing inside the ODL controller as an OSGi bundle. The App polls the text file in which the alerts generated by the IDS are written to in real-time. The sFlow packet samples from the switch are sent to the App which processes the information further. 6.1 SNORT Execution and Alert Generation Once we have the setup in place, we create alerts by carrying out certain deliberate attacks on the system. One common form of reconnaissance as a prelude to an attack is port scanning on the hosts in the system. We can use the nmap tool to do port scanning of a range of ports in a host system to identify open ports and the services running on them. This helps in exposing the possible weaknesses of the system and offer information platform to launch an attack. An example of one such nmap scan is shown in Figure-2 below where a host is scanning for open ports in the 10.0.0.1 machine.
  • 4. Copyright 2015 Page 4 of 7 Figure 2: Nmap port scan of a network node The results of the scan show there are no open ports in the host 10.0.0.1 but the very fact that a scan has happened should be good enough to trigger an alert from the IDS that there is a possible source address which is trying to do a reconnaissance activity. If the ‘scan.rules’ is enabled in the configuration file and the right IP address of HOME_NET address is populated, then SNORT should throw up an alert as in Figure-3 below: Figure 3: SNORT Alert As per the alert generated, it is clear that SNORT has been able to identify a nmap port scan trying to attempt an Information leak from the host. The alert clearly identifies the source of the reconnaissance and the victim machine. Based on the configuration, the alert is identified as a ‘Priority 2’ alert. The alert notification is parsed by the App and based on the alert priority, it is used to determine the threat mitigation action to be taken. The App gives an option to configure the blocking period corresponding to different priority alerts (with higher blocking period for higher priority alerts) for traffic emanating from the source address that is found to be generating the attack. Since addresses can be spoofed, hence it makes sense not to block a particular IP permanently. However, if the same source IP is found to be present in the alerts more than ‘n’ times (where ‘n’ is configurable), then the fraudulent IP is put in the blacklist. An address once placed in the blacklist would mean that all flows initiated from the fraudulent source would be permanently blocked till there is separate administrative action to remove it from the blacklist. Figure-4 below shows a flow that is blocked for a period of 120 seconds corresponding to a Priority 1 alert. Post the timeout period, the flow will be automatically removed and normal traffic flow can resume. Figure 4: New Flow based on SNORT Alert The above example was an illustration of how alerts are generated and corresponding action is taken by the App in response to the alerts generated. Similar alerts will be generated when any other attack is detected and the alert file generation will be similar to the one shown above. 6.2 Maintaining a White List / Blacklist of IP Address (es) In any network, port scanning activity or sending ICMP echo messages are part of the regular management messages in the network. However, attacks may actually be disguised in garb of such regular management messages. In this implementation, the App takes it upon itself to distinguish between genuine management messages from the fraudulent activities initiated by attackers in the network. To this end, its necessary that the App maintains a White List of IP addresses which are recognized as genuine IP addresses and activities matching the alert notifications from such IP addresses are not classified as harmful. The White List of IP addresses is populated in a file that it read by the App when it starts up. The App also exposes an API that can be used to dynamically add/modify/remove IP addresses from the White List as required. Just as a White List, we also have a Black list of IP addresses that are identified to be fraudulent IPs. This Blacklisted IP addresses are read by the App during initialization. If there are any addition/modification/deletion to the list at run time, this can be communicated via an API exposed from the App to do the re- configuration. The App, therefore, makes sure that the controller blocks any flows emanating from such fraudulent, black-listed IP address upfront. 6.3 Detecting Topology Changes in the Network If a new network segment is created, then that should be automatically detected and probed whether the newly added network is protected by IDS or not. The solution has an agent associated with IDS that can share the Host IP subnets that are monitored and the list of IDS rules enabled. The App keeps monitoring the network topology for any additions / modifications to the network via the topology manager and host tracker services in ODL. If any of the subnets added to the network under the purview of the controller is not protected, then intimation is sent to intimate the administrator for corrective measures. The network Admin can spawn a new VM and run IDS that can scan the new network segment for possible threats. The alerts
  • 5. Copyright 2015 Page 5 of 7 generated from the new IDS needs to be sent to the App for future threat mitigation. 6.4 Dynamic Tuning of the IDS - Add/Mod/Del IDS Rules IDS tuning is an absolute necessity for minimizing generation of False positives. The tuning is a tedious task since it has to be tuned according to the network configuration. The tuning of the IDS is a continuous activity if the network configuration changes dynamically over a period of time. The App takes the help of the IDS agent to check whether the snort.conf has configured for the Host Network ($HOME_NET) to ‘any’. The network subnet to be protected against needs to be assigned to the Host Network variable to generate definitive information as per requirement. The App interacts with an IDS agent to create new customized rules, as required by the network. The agent interacts with the IDS to gather information and provides the same back to the App controlling the threat mitigation. The new rule addition may be a requirement apart from the rules that come pre-configured with Snort. The IDS agent is also instrumental in enabling or disabling rules in the snort configuration as per the network settings. The configuration file, may need to dynamically enable or disable rules based on the traffic pattern in the network. If correct set of rules are not enabled, the attack detection will not happen. Conversely, if unnecessary rules are added, it increases the processing time of the IDS. Hence, rule optimization is necessary for efficient processing of the IDS. Additionally, the App applies a hook on the controller to check for the type of traffic packets exchanged in the network. If any traffic is generated which does not conform to the protocol types allowed, then the App introduces flows to block traffic from the source matching the protocol - so that the traffic flow with that protocol is blocked immediately. If the protocol is to be later allowed, then the flow matching the protocol type needs to be removed by the App. Conversely, we may have rules pre-configured to prevent traffic for a particular protocol. But if any such traffic is generated as per genuine network requirement, then the source IP address for such traffic should be placed in the ‘White List’ to notify the App that such type of traffic should be allowed and legal. The App will then detect such traffic and will desist from introducing a blocking flow for the source IP even though it matches a blocked protocol. Hence, the hosts whose IP address is included in the white listed IP address, have the liberty to carry any activity which may be blocked for other hosts. 6.5 sFlow Based Detection of Flood Attacks in Internal Network We have briefly discussed the sFlow integration in earlier sections. In this section, we give a detailed outline of sFlow integration with our application and the logic around it. sFlow is sampled flow [3]. It used for random sampling of packets in high speed networks. sFlow has two components - one is the sFlow agent and the other is the sFlow collector. There are several device manufacturers that support sFlow which also includes the Open VSwitch (which is used in mininet setup). The sFlow agent is inbuilt in the device and needs to be configured when bringing up the switch. However, there needs to be a separate ‘collector’ which listens to the sFlow sample packets and utilizes the sample flow to make some intelligent judgment regarding the traffic flow in the network. In the concept implementation, we have configured for sFlow in the switches which cater to traffic internal to the network. This is to track the traffic that does not flow through the gateway switch. Hence, any violations by traffic packets in the internal network will not be detected by the security devices at the perimeter. This leaves the internal network open to attacks. It’s precisely for this reason that we configure sFlow and monitor the network. A flow in the sFlow context is defined as traffic from the Data Source to the Data destination. For example, traffic flow from Data Source - say IP_1 to Data Destination - say IP_2, is treated as one flow while traffic from another Data Source - say IP_3 to IP_2, is treated as a separate flow. A flow sample contains the details of the Layer-2 to Layer-4 details of the packet exchanged along with the size of the packet that is exchanged. In our solution, the App which is running as an OSGi bundle within the controller, acts as the collector of the sFlow packets. It listens to the port ‘::6343’ for packets and their information sampled by sFlow is sent to the collector. The sFlow sampling logic is based on the concept of probability based sampling. If there is large number of packets sampled, the packets from different flows in the network gives a deterministic representation of the actual packet count corresponding to each flow. If there are a total number of packets, N, and a total number of samples, n, of which a certain number belong to a particular flow, f, then the number of frames in the flow, Nf, can be estimated using the equation: Nf = (f/n) * N [3]. According to statistical calculations, the percentage error corresponding to a flow is depicted as: %error ≤ K * √(1/f) (where K is a constant) [3] Hence, as ‘f’ increases, the error possibility corresponding to a flow goes down exponentially. While configuring for sFlow, there needs to be sampling rate and polling frequency mentioned. The sampling rate determines how frequently the flows are sampled. For example, if the sampling rate is configured as 10, then it implies that one out of 10 flow packets will be sampled and sent to the collector. Based on the traffic rate flowing through a switch, the sFlow sampling needs to be configured accordingly. This should ensure that sufficiently large number of packets is sampled and calculations based on such flows mirrors the actual packet flows in the switch. Figure-5 below shows a flow sample collected and its contents in the Raw packet header:
  • 6. Copyright 2015 Page 6 of 7 Figure 5: Raw packet header from Flow sample On the other hand, the counter samples are used to obtain the number of input and output packets that have been exchanged on an interface over a period of time. The Figure-6 below shows a counter sample collection: Figure 6: Counter Sample contents The sFlow configuration on the switch needs to ensure that a large number of samples is collected so that it is reflective of the actual traffic flow in the network. The logic applied in the App (collector) is to average the data packet sample information over a period of time (with continuously changing reference window) so that we are able to deduce the average bandwidth utilization over different interfaces. The bandwidth utilization in each interface is calculated and compared against the threshold bandwidth utilization levels for each interface. If the bandwidth utilization by a flow is found to be exceeding the threshold value, then the source address of the flow is blocked. Exceeding the threshold bandwidth is assumed to be a case of flooding the destination host interface. The period of blocking of the interface is made configurable for the network admin to decide. Also, the period of calculation of the average bandwidth utilization on an interface can be configured by the administrator user. The App also takes care to calculate the average bandwidth utilization levels over a period of time to ensure that sudden spikes in usage even out and does not cause unnecessary blockage of the source. 7. SOLUTION DIFFERENTIATORS Having outlined the implementation strategy, its important to highlight the solution differentiators and the Capex and Opex benefits associated with it. 7.1 Pro-active Mitigation with Zero Impact on Traffic One of the major drawbacks of deploying an in-band security appliance in the network is that it affects the traffic throughput of the interface to which it is attached, thereby becoming a bottleneck for the network that is protected. Conversely, there is an obvious advantage as well since in-band devices (usually placed at the gateway of a network) can match the packet signature of the incoming packets on the wire before it can intrude the network. Security devices deployed in-band and having threat mitigation capability (eg. IPS), offer a pro-active mechanism of protecting the network whereas out-of-band devices (eg. IDS) offer re-active security. In the IDPS solution implementation, the SDN controller controls the entire network. Hence, the intrusion prevention role is played out by the controller although threat detection is made by IDS which operates in promiscuous mode. So not only is our security mechanism robust in terms of providing pro-active mitigation but additionally provides non-intrusive threat detection capability as well. This is the uniqueness of the solution that is sought to be highlighted in this paper. 7.2 Low cost offering for ODL (CAPEX Saving) With software based IDS, no proprietary hardware and automation of security configuration, the implemented solution provides a low cost offering with ease of Deployment in SDN. Presently there is only an offering by Radware which is named ‘Defense4all’ to secure SDN networks[4]. However, the solution requires use of a proprietary “Attack Mitigation System” which does the scrubbing of the packets that are identified to be possible perpetrators of an attack. Additionally, it needs to be mentioned that ‘Defense4all’ is a DoS mitigation solution only. The IDPS application, in conjunction with IDS, offers protection against reconnaissance attempts, malware detection, DDoS attacks and many more. Also the system offers ease of deployment with automation of the IDS tuning based on network traffic patterns. Logs and alert information thrown by IDS coupled with statistical information collection at the controller provides for actionable security intelligence to reduce time for collating and correlating threat information. With automated threat analysis, this low cost solution is a replacement for Security Information and Event Management tool, thereby providing huge potential for CAPEX savings.
  • 7. Copyright 2015 Page 7 of 7 7.3 Ease of Deployment (OPEX Saving) The IDPS application is developed as a plugin that can be easily installed within the ODL framework as an OSGi bundle. The application is agnostic to the security device used in the network. It can be integrated with any security appliance with customization of the module that reads the alert file for processing of the threats from the network. Additionally, the automated IDS tuning capability provided by the IDS agent also aids in the deployment process of the IDS that helps to keep it in sync with the network traffic pattern and weeds out unnecessary processing of signatures that are not required. This helps in speeding up the processing of the signature based attack detection at the IDS. With an obvious ease of deployment, actionable intelligence and most importantly, proactive threat mitigation, the solution helps in replicating a security operations center which brings in the opportunity of huge OPEX savings. 8. CONCLUSION The implementation approach tries to encompass the security in SDN based networks. We have already discussed in depth about the rationale of using an IDS vis-à-vis an IPS for building our solution and the benefits to be reaped out of it. However, it must be mentioned that the system still lacks the defense against single packet attacks that will have already impacted the network before its detected. These attacks might be protected against by using a Host IDS or any end-point security that is deemed suitable. Also, there is a bound to be a time lag between generation of the alerts and their processing since the alerts need to be polled by the application before they are processed. However, this time lag is not significant to be a cause of concern since the polling of the alert information is frequent and ensures that the packet is picked up for processing almost immediately as the alert is generated. Also, it must be kept in mind that reasonably well tuned IDS will not throw up an awful lot of alerts. Hence, the system is not much worse off in terms of attack response but significantly better off in terms of removing the network bottleneck that is brought about by an IPS. With the future of networking bending towards SDN and virtualization, it’s important to have a multi-pronged approach to solve the challenges thrown by the new architecture. This implementation has been an attempt towards that end to combine the features of traditional IDS with an application to convert a re- active design to a proactive one through automation and also bring in the flavor of sampling based analytics to aid in protecting the network. The solution is by no means touted as a comprehensive solution for SDN networks. As already mentioned earlier, securing SDN networks has a wide range of challenges which include (but not limiting to) securing of the North and South Bound interfaces of the controller, securing the controller itself from external attacks, securing the VNFs, establishing trust between the elements of the SDN network and many more. The paper wishes to draw the attention of the reader towards a prospective seeded solution that could be handy towards building a holistic security design for SDN networks. The proposed solution is intended to be the initial spade work towards a larger goal of a fool-proof security of the entire SDN eco-system. 9. ABBREVIATIONS API – Application Programming Interface DDoS – Distributed Denial of Service DoS – Denial of Service IDS - Intrusion Detection System IDPS – Intrusion Detection and Prevention System IPS - Intrusion Prevention System NFV - Network Function Virtualisation OVS – Open VSwtch ODL - Open Day Light OSGi - Open Service Gateway initiative SDN - Software Defined Networking SIEM - Security Information and Event Management SOC - Security Operations Center VM - Virtual Machine 10. REFERENCES [1] http://blogs.gartner.com/andrew- lerner/2014/12/08/predicting-sdn-adoption/. [2] http://newsroom.juniper.net/press-releases/new-juniper- networks-study-finds-u-s-companies-sp-nyse-jnpr-1134411 [3] http://www.sFlow.org/sFlow_version_5.txt. http://www.sFlow.org/packetSamplingBasics [4] https://wiki.opendaylight.org/view/Defense4All:Tutorial