AUTHOR: Vladimir ‘Val’ Taft, Business Solutions Architect
Sandy was a reminder. As Hurricane Sandy ravaged the Eastern seaboard of the United States, it brought devastation and human suffering across 15 states. The storm's destruction flickered across the digital world as well, causing flooding in some of the major data centers, including those in lower Manhattan. As a result, it brought down a number of high-profile web sites and critical applications.
Speaking of extreme situations, I vividly remember working on a project to plan the relocation of a data center out of what had been and, with the recent events in the Middle East, became a war-zone location again. Relocation of 24x7 business critical applications (BCA), not surprisingly, was squarely in the path of this time-critical project. The pressure was real!
These are extreme situations, but even data center move is akin to a natural disaster in that it necessitates a relocation of an application to a different site, albeit at a known date. Business critical applications are called that for a reason – they are critical to the business. So even a regular data center move, for example due to a lease expiration, poses many questions for the CIO: How easy it will be to relocate our business critical applications? How long will it take? Will our IP addresses change? Will it break our legacy applications? What type of downtime will be required? How much new equipment remains within our existing service-level agreements (SLA) and service-level objectives (SLO) commitments?
Deploying an application as a set of virtual machines and storage virtualization greatly simplifies this process and streamlines application relocation. The IT industry seems to now be in agreement that virtualized applications are easier to deploy, run, scale and relocate. For these and other good reasons it is only a matter of time until BCA virtualization becomes a recognized best practice for contemporary IT.
There are, however, a number of barriers hindering BCA virtualization as they:
- Typically constitute a group of integrated, interdependent applications (known as an application affinity group)
- May contain legacy applications, thus may be susceptible to breaking down with infrastructure and/or IP addressing changes
- Need a mature platform as a service (PaaS) infrastructure as relocating a BCA onto an unreliable platform may and usually does create a big problem
- Tend to have stringent SLOs and highly customized infrastructure, thus require cumbersome infrastructure- and application-specific disaster recovery (DR) methods
- Require multiple non-production environments to support elaborate software development lifecycle (SDLC), which require changes to be done in technical and organizational synchrony
- And may warrant building an enterprise-wide consensus as its implementation plans and schedules require a wider, more elaborate coordination.
When the CIO faces these obstacles, it is not uncommon to do nothing, that is, continue business as usual. This problem avoidance is not a good strategy however, as it leads to:
- Sub-optimal workload sourcing
- Longer outages of the business critical applications
- And higher costs for relocation and recovery.
How real are the technical risks in BCA virtualization? It took the IT industry a relatively short time to iron out the engineering risks – and by now virtualization of the major IT infrastructure components has become a routine endeavor. The typical low-hanging fruit have been the infrastructure components owned by the corporate IT and not directly visible to the end users. This includes: directory servers; print servers; communications and security appliances; and JEE platforms including WebSphere, Weblogic, and JBoss. Over the past few years, virtualization of the enterprise-class databases: virtualization of MS SQL Server and Oracle, including its clustered variant - Oracle Real Applications Cluster (RAC), has been done and vague supportability concerns have been demystified. So, in actuality, BCA virtualization is a number of problems each of which has been sold. In fact, many of them have to be tackled at the same time, which may be daunting in some cases.
Taking into account the wide scope of technical complexity and potential organizational issues, BCA transformation needs to be approached as an enterprise-wide program. Here are just some of the critical prerequisites the CIO will need to address:
- A comprehensive business case analysis must be done in the enterprise-wide context that includes facilities, business continuity (BC), DR and strategic IT planning
- A design must be completed and a budget defined for a set of non-production environments for the BCA virtualization and a highly available production PaaS
- A credible plan must be defined to guide the implementation, rollout and cut-over to production.
To get your BCA virtualization initiative off the ground, you will need to:
- Draft a high-level plan for your BCA virtualization initiative and start the process for the inclusion of BCA virtualization in the strategic IT plan
- Review your business case with the corporate executive management, obtain preliminary budget approvals required to get started
- Include your BCA virtualization initiative in the strategic IT plan
- Negotiate a realistic timeframe for completion (18 - 24 months is desirable)
- Expedite the planning phase and building consensus by having your business case analysis for the BCA virtualization completed early in this process
Technical complexity of BCA virtualization and cost comparison with the business-as-usual baseline is not a trivial exercise. It requires understanding of all facets of IT, including the applications portfolio, technical infrastructure, facilities, DR planning, security, organizational readiness, and last but not least - value engineering. As CIO, it is important to prove to yourself and to your IT organization that for your enterprise, BCA virtualization will indeed lower the TCO for enterprise applications, streamline DR, and help data center mobility.
Which brings us back to planning a data center move and applications relocation. Readiness for common situations like data center relocations, and even more so – mobility in the face of natural disasters like Hurricane Sandy or earthquakes as well as those man-made – call for BCA virtualization . And, yes - it does help to be prepared.
VMware Accelerate can help with expeditious business case analysis and the blueprints for successful IT initiatives, thus shorten the critical path to the enterprise-wide BCA virtualization and application relocation readiness.
Visit our Web site to learn more about our offerings, or reach out to us today at: firstname.lastname@example.org for more information