Virtualization: Management and Implementation

March 7th, 2007

Many organizations are embracing virtualization technologies (VMs) and are actively moving toward large-scale implementations. This trend is still early in its adoption cycle, and analysts predict that the total VMs deployed in the next 2-3 years will be orders of magnitude greater than what was done in 2006. IDC estimates that 500,000 VMs will be shipped this year, but that the number of VMs shipped by 2009 will be a projected 1.9 billion. 

Large-scale corporate adoption is proving to be difficult

Difficulties in carrying forward the initial success of virtualization projects to the enterprise level usually lies in the organizational structure in which they are undertaken. Early projects are often initiated on non-production systems. While this approach demonstrates the technical effectiveness of the virtualization technology, it is not subject to the same challenges and constraints that exist in enterprise-wide undertakings, which typically include: 

  • Asset Identification – not knowing what servers you have, their locations, owners, functional roles, applications, etc.
  • Inconsistent IT Operations – Service Management and Service Delivery, as defined by ITIL © as well as considerations for security and access
  • Technical Constraints – network connectivity requirements, storage requirements, controllers, peripherals, UPS requirements, etc.
  • Business Constraints – availability targets, maintenance windows, application owners, compliance restrictions, Disaster Recovery relationships, etc.
  • Workload Patterns – application loads, resource distributions, overall utilization versus capacity, metrics for chargeback, etc. 

The ultimate goal of any virtualization initiative is to establish the best possible virtualization roadmap, or “transfer function”, that is optimized for corporate goals, virtualization strategies, business constraints and IT infrastructure requirements. To generate an accurate roadmap, in a reliable manner that minimizes risk and maintains reliability, it is critical to: 

  • identify all systems and applications within scope in the target environments
  • audit these systems to enable scrutiny at a deep level – what really has been installed, patch levels, etc.
  • qualify systems based on technical and business constraints
  • profile systems to determine utilization patterns – performance and capacity managementoptimize against all of these areas to determine the best transformation 

Analyzing Business Constraints

Business constraint analysis deals with a more subjective but equally important realm of information. Because of this subjectivity the proper analysis of business constraints are often overlooked in initial virtualization implementations. VM vendors do not generally touch upon this more critical aspect of server consolidation. 

Large-scale virtualization initiatives and analysis, must revolve around business constraints. This often stems from the fact that business constraints cannot necessarily be removed by throwing money at them, as opposed to technical constraints, which can often be removed by a system upgrade (the cost of which is modeled in the analysis rules in a “remediation cost” field). If you are prohibited from moving an application between locations, from combining certain applications on the same networks, or combining highly-available applications on the same server as non-critical ones, then no amount of money or effort will make these constraints go away.