Mission Critical Application Virtualization

RSS

Virtualization has become a key IT infrastructure technology. Its ability to facilitate server consolidation increases your hardware ROI while simultaneously reducing your management and infrastructure requirements. Single purpose servers typically average about 10-15% utilization. Virtualization and server consolidation can drive that utilization rate up to 75-85% -- giving you a lot more bang for your buck. Virtualization also offers clear benefits in deployment, availability and management. Using prebuilt images you can deploy new virtual servers in a matter of minutes eliminating lengthy hardware deployment and setup times. Technologies like Live Migration provide improved availability by enabling you to perform scheduled system maintenance during normal working hours with no interruption of end-user services. This makes life easier for IT by eliminating the need to perform maintenance during off-hours and for end users as well by providing increased application availability.

The key gotcha to virtualizing mission critical applications has always been performance. By definition server consolidation means you’re running multiple workloads on a single shared hardware platform. This means you’re not going to get that same performance as if that server workload were running directly on the physical box. Recent advances in server hardware, virtualization and storage have really effectively offset the virtualization performance penalty. Today’s multi-core x64 CPUs with Second Level Address Translation (SLAT) provide much better performance and virtual machine scalability than the previous generation of hardware making it possible to achieve higher levels of server consolidation and still meet your SLAs. Enhancements in the virtualization platforms like Hyper-V and vSphere have also significantly boosted VM scalability. With the release of Windows Server 2012 Hyper-V and vSphere 5.1 both platforms are capable of supporting VMs with up to 64 virtual CPUs and 1 TB of RAM. That offers enough scalability to handle all but the most extreme workloads.

While todays CPUs and virtualization platforms offer the raw compute power to handle most virtualization workloads it’s just as important to make sure your storage systems can handle I/O intensive virtualized workloads like SQL Server, Exchange and SAP. Modern advancements in storage can both improve the performance of your mission critical virtualized applications and make them easier to manage as well. Technologies like automated storage tiering can take the guesswork out of configuring your storage by automatically moving the storage for your hot I/O intensive workloads to high performance flash storage and your cooler I/O workload to lower performance higher capacity drives. On-going monitoring allows automated tiering solutions to dynamically adjust to changing resource requirements. Other new storage technologies like all-flash storage arrays can provide huge boosts in performance for I/O bound workloads by eliminating the rotational latency that’s associated with standard disk drives.

Today there’s no real reason not to virtualize your mission critical applications. With proper planning the performance of virtualized applications can rival physical servers and the advantages like the speed to deployment, improved availability and disaster recovery will continue the push to make virtualization the standard for all IT infrastructure servers.

Discuss this Blog Entry 1

on Mar 29, 2013

I can't help but wonder if stating that "...the performance of virtualized applications can rival physical servers.." doesn't need just a bit more qualification. I would think something like "... the performance of today's virtualized applications can rival the performance of the physical servers those applications were originally designed for..." is perhaps a bit more to the point, given that most of today’s applications still date back to design cycles on x86 single-core (maybe hyperthreaded) machines with only a couple GB of memory resource availability.

No doubt though, I think we agree that disk I/O is the primary bottleneck for virtualized environments. When we talk about the benefits of server consolidation because single purpose servers are running at 10%-15% utilization (and these are those single-core/hyperthreaded or dual-core systems we bought a few years ago), it'll take a lot of virtual machines to use up the CPU cycles available on a multi-socket/multi-core server -- arguably more virtual machines that can be supported on the drives that can be physically installed in such a box.

Storage management will likely be the most important near-term consideration for organizations looking to invest in virtualization. The importance here is to select the correct storage methodology for the virtualization environment being built (there are more choices available today than there were five years ago), and to ensure that the correct management tools exist for managing both the storage infrastructure and the virtual machines utilizing those services.

Please or Register to post comments.

What's End to End IT?

Blog all about IT and EMC

Blog Archive

Sponsored Introduction Continue on to (or wait seconds) ×