With server virtualization increasingly the focal point of infrastructure design activities and as organizations gain deeper experience in operating virtualized environments, it is worth considering how its widespread adoption impacts other aspects of the IT infrastructure. One of the basic promises of server virtualization, of course, is consolidation and improved utilization, and this is commonly the case, but what, for instance, are the potential implications, if any, on storage management and efficiency?
As a case in point, common storage design practice in VMware environments involves establishing numerous shared VMFS volumes generally consisting of one storage LUN each. From a typical storage management perspective, VMFS volumes are monolithic data stores. When a virtual machine (VM) is created, its virtual disks are allocated from within the VMFS volumes as individual files. It is the responsibility of the virtual server administrators to apportion and mange these virtual disks residing within the VMFS volumes, a role that was usually performed by storage administrators in the physical server world. In effect, the provisioning and management of storage has been split into two distinct functions with a significant portion of the storage administrator's traditional role shifted to another entity. The potential risk is that, without appropriate safeguards, storage efficiencies previously achieved can inadvertently be impacted resulting in poorer utilization and inefficiently allocated storage.
Avoiding this problem is foremost a process issue. The key consideration is whether the inspection and control functions for storage that (hopefully) existed in the physical environment been effectively transitioned to the virtualized infrastructure. Here are a few items to consider:
- How will policies related to assigning applications to appropriate storage tiers be addressed when those applications are residing in VMs?
- How will storage capacity planning issues be handled and utilization monitored within VMFS volumes?
- What processes must be established to enable the proper recapturing of stranded storage as VM configurations change?
As expected, a variety of technologies are becoming available to assist in this effort. Capabilities such as thin provisioning can help to avoid over-allocation; the challenge of visibility of VM storage in fibre channel SANs can be aided with support for N-port ID virtualization; storage virtualization can enable the realignment of data to appropriate tiers; and server operational management applications can help to establish configuration standards and enforce policy-based compliance.
Server virtualization presents both opportunities and challenges. None of these issues are insurmountable, but they do represent considerations that it is wise to address during the planning stages of a consolidation effort rather than as a rework and retrofit project after the fact.
Jim Damoulakis is chief technology officer of GlassHouse Technologies Inc., a leading provider of independent storage services. He can be reached at jimd@glasshouse.com.