infoTECH Spotlight Magazine

infoTECH Magazine


October 01, 2011

Why Virtualization Has Stalled and How to Restart It

This article originally appeared in the Oct. 2011 issue of InfoTECH

In many data centers, virtualization has stalled. The culprit: complexity. Look to automated data center management tools to keep forward progress. The benefits of virtualization are well known. Virtualization has proven that it can achieve key IT objectives:

  • Time to value: Deliver systems faster – both new systems and upgrades.
  • Flexibility:  Repurpose systems more easily and rapidly as needs change.
  • Uptime/availability:  Eliminate scheduled downtime for most applications, and mitigate unscheduled downtime cost-effectively with redundant virtual systems.
  • Lower capital expenses (CAPEX): Consolidate low-utilization equipment.
  • Lower operational expenses (OPEX (News - Alert)): Reduce power consumption, streamline equipment refreshes, and speed or automate operation procedures.

Virtualization seemed poised to sweep through every element of the enterprise data center. But IDC (News - Alert) and Forrester, among others, have discovered a marked decline in rate of virtualization deployments. Their customer research demonstrates that virtualization has stalled in many data centers, leaving many systems non-virtualized. Some data centers have stopped at single rack installations.

IDC’s “Server Virtualization Tracker” report in 2010 showed that the number of workloads virtualized was stabilizing around 60 to 70 percent, but only about 10 percent of mission critical apps were virtualized. Further, the proportion of servers used for virtualization was stabilizing around 25 percent. While the benefits of virtualization have proven out for numerous small or non-critical applications, there are many more application workloads that have not been virtualized, and progress is slow.

It’s easy to understand why the pace of virtualization has slowed. The first servers to be virtualized are the ones that have low utilization, or that are simple or non-critical single-function servers. Once these are virtualized and consolidated, the remaining servers are harder. Applications that have complex networking, storage, or software, or that carry critical functions, are the last to be virtualized. Operational and troubleshooting procedures must be just as refined in the virtual environment as they have been in the physical one. Restructured virtual storage must be configured and validated. Redundancy and security strategies must be re-evaluated and adapted to virtual infrastructure where virtual machines are free to move from one physical server to another (for example, using VMware’s Vmotion).

Customers have a clear preference on how to deal with complexity: Automation.  In the Forrsights Hardware Survey of late 2010, 60 percent of data centers rated automated management of virtualized servers to be a high priority. Automated management promises to directly improve availability and cut OPEX by preventing errors and by speeding troubleshooting when problems do arise. Properly applied, automation also improves flexibility, by allowing policies to be automatically maintained during times of change in the data center.

This is especially important for progressing today’s virtualization efforts. Unlike the simple servers and applications that have already been virtualized, the remaining systems may have tiered storage across SANs, multiple IP numbers on more than one VLAN, and security that is configured into intrusion protection or switch hardware access lists. There are many things that could go wrong, both when first migrating an application to virtual infrastructure, and then while maintaining it as virtual machines move between physical servers. 

Centralized and automated configuration management, particularly for network configurations, can be a key enabler for complex virtualization. By implementing automated network management today, spanning both virtual and conventional network architecture, the virtualization of complex systems is greatly simplified. 

Virtualization increases the number of virtual network devices, such as those employed by hypervisors, storage endpoints, virtual switches, and virtual network appliances, as well as numerous virtual machines on each physical host. As a result, a single physical server may support dozens or even scores of IP addresses on multiple sub-networks. Furthermore, the mapping between IP numbers and physical devices changes constantly due to Vmotion and due to the creation and removal of virtual machines. Automation and centralized management may be the only way to remain confident that data center policies and procedures have been maintained.

Troubleshooting complex virtualized data centers is also greatly improved with automated configuration management. Yankee Group (News - Alert) estimates that 90 percent of the time spent resolving data center problems is simply identifying the source of the problem.  With the additional layers of complexity in virtual infrastructure, automated assistance in tracking down problems becomes critical.

There is still much value to gain from virtualization in today’s data center.  The advantages of time-to-value, availability, flexibility, and reduced costs are still to be gained by virtualizing the conventional servers that remain in place today.  To gain these advantages, complexity must be tamed by automated and centralized system management.  Here’s how to start:

1.     Measure your own progress.  What percentage of systems and applications use virtual infrastructure today?

2.     Identify the barriers to continued virtualization.  What sorts of complexity are present? 

3.     Find tools that master the complexity today, across both virtual and conventional systems. DDI (DNS, DHCP and IPAM) tools can be a good fit, especially when they model virtualization and can discover and map both static physical resources and dynamic virtual ones.

4.     Use automated configuration policy to reduce operational staff time while increasing system stability.  Enforce policies like access control lists, firewall rules, and load-balancing algorithms consistently regardless of whether applications are virtual or conventional. 

5.     Practice troubleshooting with automated tools.  Your time-to-resolution for problems should drop, and should be consistent whether virtual or physical systems are involved.

Automation can free human resources while reducing the risks of change. With more resources and contained risk, you can take on the virtualization challenges that are stalling deployments outside a single rack or data center today.

Steve Garrison is vice president of marketing at Infoblox (News - Alert).

TMCnet publishes expert commentary on various telecommunications, IT, call center, CRM and other technology-related topics. Are you an expert in one of these fields, and interested in having your perspective published on a site that gets several million unique visitors each month? Get in touch.

Edited by Stefania Viscusi