Virtualizing Storage for VMware Success

Server virtualization has made day-today life a whole lot simpler for systems administrators; however, the benefits have not been shared by storage administrators and backup operators, who face new challenges as the result of data encapsulation. Storage infrastructure and operational practices that were designed for physical server architectures are not well suited to meet the demands of a virtual infrastructure.
Backup and disaster recovery practices have to change radically. New light needs to be shed on the importance of data protection and recovery time objectives (RTOs). Any consolidation effort also consolidates risk. With VMware the point of risk has shifted away from servers to shared storage. A RAID failure in a virtual infrastructure is catastrophic, as it potentially impacts dozens of virtual servers and applications. Any restore operation that requires data to be copied from another form of media can take days. NetApp storage virtualization technologies solve these challenges and alleviate the burden on storage and backup teams.
In this article, I’m going to describe how NetApp can:
- Cut storage costs by half or more
- Reduce VMware RTO to minutes
- Increase efficiency to decrease soft costs
If you’ve virtualized your servers but not your storage, you’ve only done half the job. Virtualizing your storage infrastructure is critical to addressing the unique storage needs of VMware.

Cut Storage Costs in Half
The combination of rapidly growing data storage coupled with the need for greater data availability has significantly increased storage costs. NetApp directly reduces the cost of VMware storage with three distinct technologies:
- RAID-DP™ improves disk availability without doubling cost.
- FAS data deduplication eliminates redundant data in production data sets.
- Flexible volumes allow dynamic storage management without disruption.

RAI D-DPv As individual disks have increased in capacity, the odds of a double-disk failure occurring have become a mathematical certainty. With a traditional array you have to choose between reducing storage costs (RAID 5) or increasing data protection (RAID 10). NetApp RAID-DP solves this issue with a dualparity implementation that provides the protection and performance of RAID 10 at a RAID 5 price point. Released in 2003, RAID-DP is implemented on more than 95% of all NetApp storage volumes in production, so you can feel comfortable the technology is mature.

Data Deduplication
VMware provides a method of provisioning a new server from an existing server template. A template is a server image that includes operating system binaries, patches, and standard application sets. Provisioning a new virtual server by cloning a template creates an identical server that is operational within minutes.
NetApp FAS deduplication enhances VMware provisioning technology by transparently deduplicating the identical parts of each virtual server to reduce production storage footprints by 50% or more with some obtaining storage savings as high as 90%. Since virtual servers deployed from templates are identical by design, this is an ideal usage of FAS data deduplication. You can free up significant amounts of storage and delay additional storage purchases. As we’ll see later, replicated virtual servers at a DR site also inherit this reduced storage footprint automatically.

Improved Storage Utilization
Unlike traditional RAID arrays, NetApp creates resource pools of storage termed aggregates, which separate the physical realm of disks and RAID from the realm of storage virtualization and storage management. Flexible volumes (FlexVol® volumes) can serve up both NAS and SAN data and allow you to carve up your storage resources more efficiently, analogous to the way that VMware lets you divvy up your server resources. FlexVol volumes spread each storage volume across a large number of spindles to ensure performance from every volume regardless of size. Using the NetApp FlexShare™ capability you can set I/O policies in a manner similar to VMware SRM in order to shape I/O limits and ensure quality of service to a particular virtual server.

Reduce Your VMware RTO to Minutes
Backing up data from a large virtual infrastructure is hard. There are many overlapping factors beginning with the disproportionately large amount of data that is addressed by each physical ESX server, coupled with limited CPU cycles and I/O bandwidth to run backup jobs. The result is that traditional backups are impractical.