Server virtualization issues contribute to data management complexityDate: Oct 16, 2012
In tip two of his seven-part series, Data Management Institute Chairman and Toigo Partners International CEO Jon Toigo explains how server virtualization adoption influences storage capacity requirements. It’s another factor IT admins must contend with in their 'storage infrastruggle,' a phrase Jon created to describe the challenges that must be overcome when dealing with storage technology. Watch the video above or read the text below to see what he has to say about curbing virtualization issues.
Please read Toigo's entire video-tip series on data management issues
Want to keep up with data growth? Sound data management is key
Enterprise hard disk overused in today's data centers
Effective storage tiering ideologies for today's data center
What to expect when using value-add software to achieve storage efficiency
Toigo: Issues with cloud storage still exist
IT planners need to focus more on data storage issues
The unbridled demand for more data storage capacity in predominantly disk-based infrastructures has helped to make storage the most costly component in a majority of IT hardware budgets worldwide. Capacity management planning must consider a range of issues, but its primary focus is anticipating and providing (in as proactive a manner possible) the storage required by applications and operating systems. Failure to do so can result in a career limiting "disk full" error and downtime that, in this time of infrastruggle, no one can afford.
A key contributor to capacity demand growth is server virtualization. Truth be told, most firms didn't fully understand server virtualization issues, requirements or the impact of virtualization on storage in general when they first embraced the technology beginning in the mid-Aughties.
In 2011, external storage demand models from Framingham, Mass.-based IDC anticipated growth in storage capacity demand of approximately 30% annually through 2014. By analyst estimates, total installed external storage capacity would be about 46 exabytes (EB) by that year. However, IDC modified its projection in 2012 to account for storage demand in shops that had adopted server virtualization in a big way. Instead of 30% annual growth, IDC projected 300% annual growth in demand for companies making major strides in server virtualization. Instead of 46 EB, the estimated storage capacity demand for 2014 in these shops is now 168 EB.
Not to be upstaged, it would seem, Stamford, Conn.-based Gartner Inc. subsequently estimated 650% growth in capacity demand over the same period, with a projected 212 EB capacity demand -- driven largely by server virtualization -- by 2014.
Are analysts correct in their prognostications? It doesn’t matter as much as the underlying thesis: Server virtualization is helping to drive up the demand for storage capacity and storage infrastructure costs.
In some cases, increased storage capacity demand in virtual server environments is a function of the dismantling of older storage fabrics (so-called SANs) and the re-engineering of storage along direct-attached storage (DAS) models, with a significant dependency on data replication to facilitate virtual server failover. In other cases, new storage rigs must be fielded to support proprietary functionality in the server hypervisor software. VMware Inc.'s vStorage APIs for Array Integration (VAAI) provides an example of how proprietary SCSI commands have been introduced arbitrarily and without standards body approval (ANSI T10 oversees SCSI and wasn't consulted before VMware introduce VAAI primitives) by a leading server hypervisor vendor, creating problems in existing infrastructure. This situation promises to worsen as server hypervisor vendors release their proprietary storage hypervisors, creating an exclusionary hardware/software stack. In a worst-case scenario, this would lead to a requirement to field a separate storage infrastructure behind each type of server hypervisor deployed.
Server virtualization is a key driver of storage capacity demand forecasts and planners ignore it at their own risk. Ideally, the impact of virtualization and the requirements it imposes on storage capacity will be factored into strategies concerning server virtualization itself to achieve better costing data for use in evaluating return on investment. At a minimum, planners need to understand the real impact of virtualization on storage capacity so they can provision for it efficiently.