Data growth is driving the use of virtualization within data centers. The virtualization evolution from server to storage to desktop is catching on at many small-to-medium size businesses, as well as at large enterprises. Aimed at providing a better end-user and administrator experience than their physical counterparts, virtualized desktops promise lower cost of acquisition and management with a highly scalable, easy-to-deploy and fully protected environment. However, with virtualization desktop infrastructure (VDI) comes a set of new challenges. Chief among these are storage and server resource allocation and data protection and recovery.
Weathering VDI Storms
VDIs tend to see sharp rises in the demand for server and storage resources during recurring events that have high input/output (I/O) demand, such as login, logoff and shutdown, or while running antivirus scans. These daily events result in I/O "storms" that can affect the performance of the virtualized environment for up to two hours. In addition to resource management, desktop virtualization also challenges VDI developers to equip their solutions with the necessary data protection capabilities users require in the event of file loss or corruption.
A typical input/output per second (IOPS) requirement of a desktop under normal workload can be five to 10 IOPS, depending on the operating system and applications being used. However, the difference in IOPS requirement between regular desktop operations and during one of the high I/O events could be 10 times higher. Compound that with the number of virtual desktop users consolidated on dense server and storage infrastructure, and companies experience a big outburst of IOPS requirement - an I/O storm from users coming to work in the morning, a logoff or shutdown storm when they leave in the evening or a combination of both during lunch time. Each one of these events can take anywhere between 30 minutes to 2 hours; if the environment is not properly sized to deal with these storms, users could experience a significantly degraded level of service.
Mitigating the impact of I/O storms typically requires sizing the environment for the worst-case scenario. Traditionally, IOPS are sized based on the number of spindles in the storage environment. Take for example a company with 5,000 employees. If every employee requires 10 IOPS for normal operations, that means the storage infrastructure has to deliver 50,000 IOPS to support them. If IT administrators consider that during morning hours there is a maximum number of employees, around 10%, who are trying to login at the same time - this creates a surge of IOPS that will cause the I/O storm. And let's assume that the average IOPS requirement for each one of these activities is 10 times the normal operation requirement. This means IT would need to account for 100% additional IOPS. This will vary from one organization to another, but in this example, the total required IOPS is 100,000.
If IT sizes the above environment for a (50/50) read/write I/O profile with Fibre Channel or serial attached SCSI (SAS) 15,000 rpm drives, they would need around 834 spindles (this varies according to type of drive and RAID configuration). This is a massively scalable environment with a lot of wasted capacity.
However, new technologies allow data centers to protect themselves against waste and IOPS storms with a mix of solid-state storage with high-density serial advanced technology attachment (SATA) drives. This combination addresses the IOPS and capacity requirements independently. By sizing for IOPS with solid-state disk drives (SSD) or solid-state memory arrays that deliver approximately 5,000 IOPS per drive, administrators can address the 100,000 IOPS requirement with 20 SSD drives while addressing the capacity requirements with SATA drives. This translates into huge reductions in acquisition costs and floor space requirements, which in turn leads to significant reductions in power and cooling for the data center.
Data Protection and Recovery for VDIs
There are several different levels of data protection and recovery that any VDI environment should have: system data, user data repositories and individual virtual desktops. In a VDI environment, all system data resides on shared storage resources servicing virtual server clusters, which allows for all that data to be protected with the snapshots and replication capability of the storage area network (SAN). This is not exclusive to any storage vendor, and most arrays today support both functions more or less efficiently. Once system data is protected, administrators can recover a full environment locally from snapshots or even remotely by failing over operations to a remote location to which data has been replicated. The primary outcome is simplified business continuity planning.
The second tier of protection should take care of the user data repository. In some cases, virtualized user data is redirected to a network-attached storage (NAS) environment where user data is hosted separately from system data. This approach is aimed at providing NAS protection to user data. A preferable approach is to allow customers to group such directories under the same data protection umbrella, providing persistent and consistent data protection and recovery capability.
On additional place to look for data protection is in the individual files and the method by which they are protected and recovered. With the model above, an end user has to call the help desk or the system administrator to recover a file in case of data loss or corruption. In turn, the administrator must get the individual virtual machine up from a previous snapshot or the user data repository and look for that file to recover and send to the end user. Needless to say, this can be a time-consuming process for the IT staff. And the traditional model of backing up physical desktops won't cut it here - running hundreds of backup agents simultaneously on a few virtualized servers would literally bring an infrastructure to its knees.
To improve the end user experience and relieve an already overloaded IT staff, data protection and recovery agents should be embedded in the VDI gold master image to enable automatic protection of user data. The user data folders can then be protected on a continuous data protection repository, allowing end users to browse their directory structures as they would their desktops and select a file and a time from which to recover in case of loss or corruption. End users can perform self-service file-level recovery to cut recovery time to seconds and free up the IT staff for more strategic tasks.
Virtualization is changing the IT world, and desktop virtualization in particular brings flexibility, mobility, higher availability and business continuity to desktops. Desktop virtualization is a new approach to enterprise-wide desktop deployments that provides a better end user and administrator experience than physical desktops, lowering the cost of acquisition and management while offering a highly scalable, easy-to-deploy and fully protected desktop environment. Nevertheless, this consolidation raises new challenges in terms of computer resource allocation and granular data protection and recovery processes. The market is addressing these problems with VDI-specific solutions, clearing the way for data centers to fully realize the potential of this technology.