"Configure once, deploy anywhere" is one of the most sought-after enterprise operations requirements. Large-scale IT shops want to keep the flexibility of using on-premises and cloud environments simultaneously while maintaining the monolithic custom, complex deployment workflows and operations. This session brings together several hybrid enterprise requirements and compares orchestration and deployment models in depth without a vendor pitch or a bias. This session outlines several key factors to consider from the point of view of a large-scale real IT shop executive. Since each IT shop is unique, this session compares strengths, weaknesses, opportunities, and the risks of each model and then helps participants create new hybrid orchestration and deployment options for the hybrid enterprise environments.
1. AWS Summit 2014
Orchestration and Deployment
Options for Hybrid Enterprise
Environments
Guy Ernest
Solutions Architect
@guyernest
2. What is Hybrid Cloud?
A composition of two or more distinct cloud
infrastructures that remain unique entities, but
are bound together by standardized or
proprietary technology that enables data and
application portability.
“Special Publication 800-145 - The NIST Definition of Cloud Computing” – September, 2011
3. Requisite Gartner Quote
“Nearly half of large enterprises will
have hybrid cloud deployments by
the end of 2017.”
http://www.gartner.com/newsroom/id/2599315 - October 1, 2013
4. Why Hybrid Cloud?
• All the things the cloud provides
– Agility
– Economics
– Scale
• But something gets in the way
– Compliance
– Previous investment
– Legacy workloads
– Attitudes
5. What do Enterprises Want in Hybrid?
• Ability to deploy identical stacks
• Interoperability between clouds
• Ability to leverage one provisioning framework
• Ability to leverage one operational framework
6. Hybrid Considerations
• Core Infrastructure
• Security
– Authentication and Entitlements
– Identity Management
– Data Sovereignty
• Operations and Monitoring
7. Hybrid Considerations
• Cost Containment
• Pace of Innovation
• Cloud Orchestration
• Application Deployment
• Processes and Change Management
9. Preparing Core Infrastructure
Active Directory
Network Configuration
Encryption
Back-up Appliances
Users & Access Rules
Your Private Network
HSM Appliance
Cloud back-ups
AWS Direct Connect
Your Data Center Your Cloud
10. Core Infrastructure Considerations
Driven by Business Requirements!
• Performance & Latency
• Business Continuity
• Geography
• Data Sovereignty
• Security
• And Many More…
11. Some Relevant AWS Services
• Amazon Virtual Private Cloud (VPC)
• AWS Direct Connect
• AWS Identity and Access Management (IAM)
12. Some Relevant AWS Services
• AWS CloudFormation
• VM Import / Export
• AWS Management Pack for Microsoft System Center/
VMWare vCenter
• AWS API, SDKs, and Tools
14. A Decision Framework
DIMENSION LOW MEDIUM HIGH
Organizational
Buy-In
None or
grassroots
Divisional Top Down (CIO/
CEO)
IT Capabilities In-house Limited
Partner Limited
In-house Limited
Trusted Partner
In-house Advanced
Trusted Partner(s)
IT Vision Operational Somewhat
forward thinking
Innovative and
cutting edge
AWS
Experience
None/Limited Some Extensive
18. Native Integration
Build a custom layer using API-level capabilities.
Best When:
• Have in-house development skills
• Need very fine-grained control
• Licensing costs are a big issue
19. Native Integration
DIMENSION LOW MEDIUM HIGH
Organizational
Buy-In
None or
grassroots
Divisional Top Down (CIO/
CEO)
IT Capabilities In-house Limited
Partner Limited
In-house Limited
Trusted Partner
In-house Advanced
Trusted Partner(s)
IT Vision Operational Somewhat
forward thinking
Innovative and
cutting edge
AWS
Experience
None / Limited Some Extensive
20. Native Integration - Pros
• Incorporate all services or only what you need
• Maximum flexibility
• React quickly to new features and services
• Leverage existing open-source tools
– Open Nebula
– Eucalyptus
– Netflix Asgard
– CloudStack
• No licensing fees
21. Native Integration - Cons
• Need in-house development skills
• Possible long development cycles
• Private cloud must support API-level access
• Support must come from in-house
22. New Orchestration Layer
Invest in new hybrid orchestration tools.
Best When:
• Have moderate time constraints
• Want the latest and greatest
• Have trusted partners
23. New Orchestration Layer
DIMENSION LOW MEDIUM HIGH
Organizational
Buy-In
None or
grassroots
Divisional Top Down (CIO/
CEO)
IT Capabilities In-house Limited
Partner Limited
In-house Limited
Trusted Partner
In-house Advanced
Trusted Partner(s)
IT Vision Operational Somewhat
forward thinking
Innovative and
cutting edge
AWS
Experience
None / Limited Some Extensive
24. New Orchestration Layer - Pros
• Get latest and greatest capabilities
• Multi-cloud support
• Faster than DIY
• Vendor-provided support
25. New Orchestration Layer - Cons
• Licensing costs
• Rip-and-replace legacy tools
• Maintaining feature parity with AWS
• Requires some specialized skills
26. Extend Existing Tools
Leverage existing investments in tools
Best When:
• Have aggressive time constraints
• Don’t need latest and greatest
• Have strong relationship with
existing tools vendor
27. Extend Existing Tools
DIMENSION LOW MEDIUM HIGH
Organizational
Buy-In
None or
grassroots
Divisional Top Down (CIO/
CEO)
IT Capabilities In-house Limited
Partner Limited
In-house Limited
Trusted Partner
In-house Advanced
Trusted Partner(s)
IT Vision Operational Somewhat
forward thinking
Innovative and
cutting edge
AWS
Experience
None / Limited Some Extensive
28. Extend Existing Tools - Pros
• No rip-and-replace
• Can be fastest path to hybrid
• Familiarity with tools and vendors
• Vendor-provided support
• Requires least amount of specialized skills
38. Who is Eyefreight
Background
• Initial Dutch Government investments into Innovation in Supply Chain Logistics
• 2009 Focus on Cost Management within Transportation
• 2012 Funded by 2 Dutch Private Equity firms
• 2013 Funding of US$ 14m
• 2014 Geographical presence
– European headquarters in Utrecht,
Netherlands
– US headquarters in Chicago, Illinois
39. Eyefreight, in other words
• Young company with well known customers
• SAAS solution offering freight spend
management and visibility
– Tightly integrate with mission critical ERP systems
– Communicate with hundreds of different parties
– Running optimization algorithms on high volumes of data is core
business
40. Being an Innovative Company
• Innovative companies are typically about doing
the right things
• However, doing things right is like hygiene
– as in “if you don’t do it right you will get in stinking mess”
• So… get the right guys to ‘do things right’ for
you. Think Services, not personnel.
• Only invest in your edge and competitiveness
41. A bit of background
Transportation
Planning
Execution
Monitoring
Cost
settlement
Transport
orders
Shipments Shipments
Order
Allocation
Stock orders
Sales orders
Transport orders Shipments Shipments
Shipment
assignment
Status
updates
42. Managing Mission Critical Loads
• So how do we handle “mission critical”?
– Load balancing and fail-over
– Streaming replication
– Backup and point in time recovery
– Configuration management in a massive cluster
– Security certification
• Yeah yeah yeah. (You’ve read the book)
Per customer / day
• 3000 orders
• 200 shipments
• 800 updates
• 15.000 page hits
Now think 500
customers, 62.000
users, 3 continents
43. But how to use services to do that
• Architect and design a solution that supports
cloud services from day one
– Support multi tenancy from the load balancers through all
components down to the database schema
– Many clusters with X nodes, using Y databases containing Z
schemas
– Create appliances that self configure (dynamic cluster sizing
made easy). Just add a node
– Isolate environments (customers, regions). Just add a cluster
44. And then go for scale
• Automate
– Configuration: Packer, Chef, Docker, OSGi
– Deployment: CloudFormation, local registry discovery, automatic
schema upgrades, Apache ACE/OBR
– Scaling: APM trending / hotspot discovery, dynamic cluster
sizing, aggregated logging
– Security: intrusion detection and global auditing
• Automation + capacity on demand = Flexibility
45. Scaling up; when you need it
Why?
SLA / Life cycle specific
Customer specific
Environment specific
Load specific
Economy of scale
46. Tips & Lessons learned
• There are tools and API’s: Automate your build
process to produce AMI templates
• If you need to scale, do not forget the rule of the
weakest link. In our case that is still RDS
• Balancing over HTTP is so much easier. REST!
47. Challenges (potentially new services)
• Backup and recovery services are at the level of
a database (not schema). So not necessarily at
the level of your customer
• Root cause analysis is a pain in a load balanced
cluster. You need to aggregate and correlate
your logs centrally
• APM style Metrics at the functional level
48. Direct benefits, entrepreneurial
• If you do not need to invest; then don’t
• Instant maturity levels for OPS
• Very flexible capacity; complete environments
can be created and abandoned ad-hoc
51. Getting Started
• Storage / Backups and Archive
• Development and Test
• Net New Workloads
• Disaster Recovery
• Cloud Bursting
• Migrate Legacy Workloads
54. Getting Started – Network Topology
Subnet 1
…
Subnet 2 Subnet N
Considerations
• Overlapping networks
• IP stinginess
• VPC CIDR too small
• Subnets too small
55. Getting Started – Connectivity
Considerations
• Public Internet vs. Direct Connect
• Redundancy
Customer
Data Center
DX Location
56. Getting Started – IAM
Considerations
• Identity Federation
• AWS vs. App Stack Access
• Build vs. Buy
57. What Next?
• AWS Account Team
• Trusted Partners
• Resources
– http://aws.amazon.com/architecture
– http://aws.amazon.com/enterprise
58. Thank You!
AWS EXPERT?
GET CERTIFIED!
aws.amazon.com/certification
Guy Ernest
Solutions Architect
@guyernest