Opinion: Virtualization efforts underway to offer end-to-end view

Computerworld - These days it's virtually (pun intended) impossible to have a discussion of storage issues in an organization without devoting a significant portion of time to the subject of server virtualization. The impact of virtualization technologies such as VMware on infrastructure planning and design is reminiscent of challenges encountered when storage area networks began to be introduced in the nineties. Implementing SAN technology spurred a rethinking of the traditional design assumptions of siloed server-storage design to effectively address both the opportunities and challenges associated with shared storage resources. It took several years for processes, technologies, and mindsets to evolve a relatively mature set of accepted practices.

With server virtualization, we see a similar scenario unfolding. SAN technology forced a transition from the extravagance of exclusive server access to disks and tape drives to a more constrained shared resources model, and virtualization now extends this to the servers themselves. While everyone appreciates the cost benefits - improved server utilization, smaller data center footprints, reduced power and energy requirements - managing and coordinating the additional operational complexity requires the adoption of new practices and tools.

Besides the obvious I/O-related resource sharing concerns, leveraging associated capabilities like virtual machine (VM) relocation add complexity to standard storage management functions. Traditional tasks like managing logical unit number (LUN) visibility and access, for example, demand extra planning and coordination.

Backup administrators, too, are grappling with how best to handle virtual environments, particularly with the tradeoffs associated with backing up VM images while still enabling recovery of individual files. Although solutions like VMware Consolidated Backup (VCB) are available to address issues like VM image open-file access, successful deployment also requires rethinking and revising standard backup operating procedures. Thoroughly understanding and implementing these changes, along with such necessary backup redesign considerations as how many VMs, each VCB server can support, has resulted in many sites opting, for the time being, to retain the traditional paradigm of deploying backup clients within each guest OS. Besides potentially being costly (depending on backup vendor client licensing), it also exacerbates resource contention and complexity issues since backup administrators often have no means to distinguish virtual from physical hosts.

Given the enormous demand, solutions are evolving rapidly. Storage management tools are becoming more VM aware and there are efforts underway to present a complete end-to-end view from application and VM through to the underlying server, network, and storage infrastructure. Backup vendors are revising their client licensing as well as introducing VM-integration features. Most importantly, the collective experiences of users is driving new levels of accepted practices. Ultimately, we are steadily evolving toward a true shared infrastructure resources model - the underlying economics are compelling - and hopefully soon the operational bumps in the road will be smoothed.