Virtualising the desktop infrastructure represents one of the phases in the journey to a private cloud. It features strongly as a solution for any organisation looking to rapidly provision hundreds of desktops on-the-fly while keeping hardware costs minimised. It also enables organisations to achieve:
● Scalability: Provision of virtual desktops can be easily extended to users
● Enhanced data security: Data and applications reside on secure data centres not on users' desktops
● Reduced costs: Costs associated with technology refreshes, power and cooling are lowered
● Improved manageability and administration: Application installation and management as well as security patches are centralised
The impact of VDI on storage
While the benefits of VDI may be comprehensive, not all organisations have the necessary storage infrastructures in place to reap these benefits.
Virtual desktop environments usually entail relatively high writes-to-reads ratio. Anything from massive user-logons to anti-virus updates and user sign-offs can strain the storage infrastructure. As large numbers of users boot-up their desktops, they contribute to "boot storms", with the surge in I/O volume potentially causing bottlenecks. For example, the typical Windows 7 desktop will require about 5-10 input-output operations per second (IOPS) in normal operation. During boot time however, this can increase to approximately 50-100 IOPS. When these peaks in IOPS fail to be processed quickly, latency results and it is experienced as lag-times by end-users.
Storage infrastructures which are ill-equipped to handle the enhanced workload VDI enable will not only impede optimal application performance but also risk increasing storage costs as organisations continually provision more storage to match rising workload volumes. Organisations will therefore need to relook how they approach provisioning storage for VDI deployments.
Storage bottlenecks unclogged
Not all workloads are born equal, with some requiring more frequent access to data stores than others. This provides an opportunity for organisations to design smarter storage infrastructures that hinge on matching application workloads based on their I/O levels to appropriate storage tiers. To achieve this, three approaches can be considered:
Storage tiering: EMC FAST
EMC fully automated storage tiering (FAST)-which forms part of EMC's Flash 1st Strategy-continuously monitors and differentiates between active and inactive data and directs them to the most appropriate storage tier based on policy settings. These policy settings enable administrators to control and manage the automated activity. It will typically involve moving active data to high-performance storage tiers such as enterprise flash drives (EFD) and inactive data to low-cost, high-capacity storage tiers such as SATA disk drives. EFDs can increase application performance by up to 800 percent while SATA drives lower costs by approximately 80 percent. A tiered approach therefore reduces storage latency while keeping overall storage costs contained.
Sign up for CIO Asia eNewsletters.