Deploying Private Cloud - Lessons from the Trenches

Year 2015 was somewhat a watershed year for enterprises. Many enterprises have taken concrete steps and formulated plans to uptake cloud in a full-fledged fashion. Enterprises have long been watching the cloud players and have been testing the waters the last few years. Many tested waters, started out cautiously and rightly so what with many vendors claiming out-of-the box functionality and seamless integration coupled with a hodge podge of various tools and infrastructure bits. Enterprises had to comprehend a slew of things from license restrictions to security/compliance to people/process. Here are some of the lessons learnt working with large enterprises worldwide leading to successful private cloud deployments.

#1 Start with the right goals. It is important to decide upfront the reason for putting together a private cloud. Is it business agility? Is it cost savings? This has to be crystal clear and this one key decision will drive all other decisions. It might be ok for some enterprises to just virtualize and realize the goals than going all the way to conceptualize a private cloud.

#2 Biggest Barrier is still cultural. The impetus for cloud deployments typically is top down but no matter how it comes it is crucial to take this as an opportunity to partner with the business side and engage a wider part of the company. All decisions should plug into the broader corporate structure - corporate Risk, Corporate Security , Change management, Asset Tracking, software licensing compliance and audit teams. Additionally Cloud transforms IT Departments from being custodians of technology to Service Orchestration of technology that can run internally and externally which is a big change.

#3 Service Orientation is Key. Take a “business first” Approach. Focus on business requirements and not just on cloud plumbing. Cloud is more about transformation than about just technology. Service orientation and design drives adoption and broader adoption drives down costs. IT departments should move from being a gate keeper to a Service Provider capable of analyzing applications growth patterns/ business processes and formulate strategy to provide advanced services including creating an API strategy for the organization.

#4 Service Catalog drives Adoption. Keep it simple to engage constituents and let the workload and service flexibility be your guide. Watch out for image sprawl and/or template sprawl. Self-service should be baked in from ground up and should be complete end-to-end without which will create the wow factor for adoption. This should be approached as a New Product Launch and with it comes all the usual controlled studies, user studies, product marketing all internally targeted.

#5 Reengineers Apps for the Cloud. Applications need to be categorized into different buckets - Legacy Apps, Webapps using object storage etc. Review Business Continuity and Disaster Recovery Requirements. Differentiate web scale apps vs Consumer facing Apps vs large scale traditional systems of record. Anticipate Change and be cognizant of Application Roadmaps. Once the nature of apps is comprehended, then put together a cloud migration model - Lift and Shift, Partition, Refactor or Re-write applications. Above all make developers love you by getting their mind share by exposing them early on to your work.

#6 Take a staged incremental approach. Start with small isolated use cases with few dependencies and have a governance model in place. Combine effect of scale, demand diversification and multi-tenancy wherever possible. Effectively use Public cloud as testing ground for Private cloud Deployment – this can help identify several things that you may not have thought about or can anticipate.

#7 Integration with legacy and existing systems. Ensure integration with legacy apps and other systems and make provision for Data Federation, data Movement and Data Governance. Build for Public Cloud Integration even if there is not an immediate need as trust me, it will come up sooner that you think.

#8 Automate and Orchestrate. One of the things that enterprises will realize soon is things are not as seamless, as “out-of-the” box or integrated as what vendors state. Be prepared to Buy or Build Management software – provisioning, configuration, monitor security, performance and availability. You should have a cloud dashboard that can provide a single pane of control. Basic Metering and Billing, Chargeback and Reporting Capabilities, Monitoring Operations and diagnostic Capabilities should be in place. Be open to take up open source tools which can be effectively integrated within your environments. Automate beyond “infrastructure as code” – without this, project may be successful but will lack flexibility and scalability.

#9 Cloud Devops provides sustenance. Devops Principles and Practices provide foundational capabilities for an Adaptive IT and incorporate right upfront Principles of Continuous Improvement. Overhaul of IT Service Model should be undertaken to uptake the cloud paradim – Design, Develop, Ops and Support is critical to go with the cloud infrastructure. Continuous Integration (CI) and Continuous Deployment (CD) with Application Lifecycle Management (ALM) can accelerate ROI on cloud adoption.

#10 Address Cloud Skills. Private Cloud is not just about provisioning VMs and Storage but calls for higher order orchestration, automation and lifecycle management skills. Develop skills organically and augment with external skills. Lack of appropriate skills typically creates delays and oversights.

We are at a point of inflection with Private Cloud adoptions. You can expect more intense rounds of innovation from cloud vendors addressing the needs of integration, monitoring, compliance, more end-to-end automation and lastly light weight containerization.