This paper focuses on a specific real-world use case for SDN - the Software-Defined Data Center. It provides Avaya’s perspective on the characteristics of the Software-Defined Data Center and the value of its Fabric Connect technology as the foundation for this solution. It also talks about how combining Avaya Fabric Connect with open-source cloud orchestration capabilities (that are being defined by OpenStack) can enable a graceful migration to the Software-Defined Data Center.
Rising Above_ Dubai Floods and the Fortitude of Dubai International Airport.pdf
Avaya Fabric Connect: The Right Foundation for the Software-Defined Data Center
1. Avaya Fabric Connect: The
Right Foundation for the
Software-Defined Data Center
This paper focuses on a specific real-world use case for SDN - the
Software-Defined Data Center. It provides Avaya’s perspective on
the characteristics of the Software-Defined Data Center and the
value of its Fabric Connect technology as the foundation for this
solution. It also talks about how combining Avaya Fabric Connect
with open-source cloud orchestration capabilities (that are being
defined by OpenStack) can enable a graceful migration to the
Software-Defined Data Center.
What is the Software-
Defined Data Center?
According to Webopedia, Software-
Defined Data Center (SDDC) is the phrase
used to refer to a Data Center in which all
the infrastructure components are
virtualized, packaged, and ultimately
delivered as a service. Control of
resources is automated by software –
meaning that hardware configuration is
maintained through orchestrated software
systems.
The SDDC is considered by many to be the
next step in the evolution of server
virtualization and cloud computing. As
components within the Data Center
(server, storage, networking) are
virtualized, the end game of the SDDC is
to combine, customize, and commission
resources from the server, storage and
networking pools to ensure that
applications or services meet the capacity,
availability and response time SLAs that
an agile business requires. The desire is to
deliver much more IT flexibility, efficiency
and agility in an organization’s business
processes so that IT Operational teams
can turn up new applications faster and be
much more responsive to their user and or
customer base.
Evolution to the Software-
Defined Data Center
Avaya offers solutions that help
enterprises and service providers prepare
for the transition to the SDDC. Many Data
Centers that have deployed server and
storage virtualization already are now
focusing on network virtualization. This
new focus will result in significant benefits
by addressing what is currently the most
complex portion of turning up and fine-
tuning applications: coordination and
orchestration with the network.
The challenge with today’s
networks
Today’s server environment is dynamic.
New virtual machines can be spun-up in
minutes and can migrate between physical
hosts on-demand or dynamically. The
network remains relatively static
however, and a painstaking, error-prone
provisioning process to provide network
connectivity for applications is too often
required. To facilitate the movement of
applications between servers, Layer 2
VLAN extensions, both within the Data
Center and across the backbone
infrastructure between Data Centers, are
avaya.com | 1
Today many point to
software-defined
networking (SDN) as
the next big thing in
networking. But
what is it really?
Initially it was closely
associated with
OpenFlow which
provides separation
between control and
data plane so that
each could be
independently
managed. But now,
sparked by new
technology options,
the definition is
broadening with
SDN being
positioned as a
means for delivering
increased agility and
automation to IT
Operational teams.
2. required to provide a robust and
transparent connectivity service.
In today’s traditional LAN/WAN
design, the extension of VLANs and
their propagation within Data Centers
can prove challenging. Ensuring all
redundant links, in addition to
Switches, are properly configured can
be a time-consuming operation and
can introduce significant risk due to
the need to regularly administer
configuration of crucial core devices.
Furthermore, initiatives such as video,
BYOD, and Big Data are putting
additional pressure on today’s
networks and are causing many IT
and network operations teams to
reevaluate how they might evolve
their networks.
Network virtualization provides a
more dynamic and agile network
environment and - when the right
approach is taken - can provide
benefits to the network that are as
profound as those experienced by
applications when server
virtualization technology emerged.
Not all Network
Virtualization
Technologies are Equal
The concept of an Ethernet Fabric is
seen as a critical next-generation
network virtualization technology.
Ethernet Fabrics allow business to
replace obsolete Spanning Tree
Protocol-based networks; this
increases network efficiency through
genuine load-sharing across multiple
active links and increased availability
by setting up Layer 2 connections
between servers in order to facilitate
the migration of Virtual Machines.
Avaya’s dynamic, real-time, service-
based Fabric Connect technology is
the most advanced Ethernet Fabric
solution on the market today. Going
beyond simple Spanning Tree
replacement, it delivers the full
breadth of desired integrated services
including Layer 2 virtualized services,
Layer 3 virtualized services (with
multiple Virtual Routing and
Forwarding instances), and fully
optimized routing and integrated IP
Multicast support.
Applicable to single and multi-tenant
Data Centers, Fabric Connect enables
businesses to gradually migrate away
from a host of legacy technologies
(such as STP, OSPF, RIP, and PIM) and
to enable all services with a single
technology. Powered by edge-only
provisioning, Fabric Connect allows
new services or changes to services
to be implemented and also
eliminates the legacy requirement of
end-to-end provisioning, including
the core, to facilitate much greater
ease and agility. It also makes the
physical topology of the network
irrelevant by allowing any logical
topology to be built, whether it is
Layer 2, Layer 3, or a combination of
the two - to deliver new levels of
flexibility for network operators.
Eliminating complex protocol overlays
not only simplifies the design and
ongoing management of the network
but can also speed up network
re-convergence times dramatically.
Customers who have already
implemented Avaya’s Fabric Connect
are experiencing recovery times of
less than 50 milliseconds network-
wide for link or node failures which
represents a vast improvement over
large OSPF routed cores and massive
improvement when compared to
legacy solutions that relied upon PIM-
based Multicast.
Vendor lock-in is a common concern
for customers who are evaluating new
technologies and unfortunately most
vendor implementations of an
Ethernet Fabric technology are
proprietary. Avaya’s Fabric Connect,
however, is based on the Shortest
Path Bridging protocol – one of the
only technologies ever to be
standardized jointly by the IEEE and
the IETF. Being standards-based
means that Avaya Fabric Connect can
interoperate with the massive
installed-base of existing Ethernet
products and thereby allow
customers to deploy the technology
in areas where they need it the most
without having to do a full rip and
replace. Avaya Fabric Connect can
be, and often is, deployed alongside
legacy technologies – running as
‘ships-in-the-night’ – allowing services
to be migrated selectively without
impact.
Avaya is committed to providing an
open approach for the foundation of
the Software-Defined Data Center
and has been actively participating
with other vendors to demonstrate
Shortest Path Bridging
interoperability. An interoperability
test was conducted at Interop 2013 in
Las Vegas with major industry
vendors Alcatel Lucent, HP, and
Spirent.
2 | avaya.com
Comparison of Ethernet Fabric technologies
3. avaya.com | 3
Critical to the Software-Defined Data
Center is the ability to move Virtual
Machines between IP subnets within
the same Data Center (and even
between geographically dispersed
Data Centers) with new levels of ease
and flexibility. A key benefit of Avaya
Fabric Connect is that it offers
unrestricted VM mobility by
leveraging a single technology that
can extend seamlessly across
distance. Virtual Machines can
migrate on-demand or dynamically
between physical servers located in
geographically dispersed Data
Centers as easily as between servers
located in the same rack. This is
achieved without any of the complex
protocol stitching commonly
practiced by competitive Ethernet
Fabric technologies.
It is important to consider scale.
While the Shortest Path Bridging
standard enables scaling up to 16
million unique service instances,
competitive Ethernet Fabric
technologies are typically based on
VLAN virtualization which limits them
to the 4096 ceiling. This is clearly not
adequate considering that many large
Data Center environments are already
experiencing VLAN exhaustion and
are looking for a network
virtualization technology that can
solve this.
Finally, while many Fabric
technologies are designed exclusively
as Data Center technologies, Avaya
Fabric Connect extends network-
wide, providing a single service
delivery model end-to-end. Services
can be deployed via simple end-point
provisioning where servers attach and
where users attach. Users can request
access to the network and be placed
dynamically into the appropriate
Layer 2 or Layer 3 virtualized network
according to predefined policies.
Virtualized networks can be
dynamically created and can reflect
anything from separating different
departments in a traditional multi-
tenant environment to separating
different types of users (wireless
guests, executive access) and
isolating traffic types for security
and/or regulatory compliance (i.e.
banking transactions for PCI DSS
compliance, medical imaging devices
in a hospital). Network access is
completely automated – users and
devices are grouped into the
appropriate virtualized network
based on their credentials.
In summary, Avaya Fabric Connect is
unique in the market and offers a new,
innovative way to build, deploy and
manage networks that deliver new
levels of simplicity, agility and uptime.
The Value of OpenStack
Cloud Software
For a true Software-Defined Data
Center, being able to orchestrate,
coordinate, and allocate resources
from server, storage and networking
pools is critical to ensuring that
applications or services meet
capacity, availability and response
time SLAs required by business.
OpenStack is an open source cloud
operating system created by NASA
and Rackspace that simplifies
creation and movement of
applications and virtual machines in a
public or private cloud. Traditional
methods of configuring a network
and storage and virtualized servers
could take months and involve several
complicated independent steps.
OpenStack provides a control layer
that sits above all the virtualized
resources within the Data Center and
abstracts administrators and users
from the underlying components
which allows them to orchestrate
those resources as a service through a
set of APIs and a common dashboard.
From the integrated dashboard,
administrators can spin up virtual
machines, assign storage, and
configure networks through simple
point-and-click workflows. When
operationalizing new applications,
this technology makes operations
consistent across multiple
technologies and offers increased
agility and simplicity, especially in
multi-vendor environments where
multiple hypervisors, storage
technologies, and networking devices
may be present.
Within OpenStack there are several
modules that manage each resource
individually. Specifically, the
OpenStack ‘Nova’ interface
orchestrates virtual machine
resources, the ‘Cinder’ and ‘Swift’
interfaces orchestrate block and
object storage respectively and
‘Neutron’ interconnects everything by
coordinating networking services and
appliances. Typically this is brought
together under the umbrella of
OpenStack’s ‘Horizon’ graphical
orchestration interface.