Virtualization and All-Flash Arrays, a Must Have?

Flash Virtualization Myth #1 – All-Flash Arrays are a “must” for Virtualization

Watch On Demand

There are a lot of myths surrounding the use of flash and virtualization, in our on demand webinar we expose the top 4. One of those myths is that All-Flash arrays (AFA) are a must-have when architecting a virtual infrastructure. It is true that AFAs can bring a number of benefits to the virtual environment including increased virtual machine density, more consistent performance and impressive results from deduplication. But are AFAs a must-have for virtual environments, or can hybrid systems deliver many of the same benefits?

The Performance Myth of All-Flash Arrays

The fact that AFAs deliver impressive performance is certainly not a myth. Even “slow” systems generate hundreds of thousands of IOPS. Where things get a little fuzzy is how much of this potential performance does the typical virtual environment need? The reality is that most virtual machines (VMs) don’t need anything close to flash performance, for them hard disk drive (HDD) IOPS is perfectly acceptable. Using a combination of flash and HDD is more appropriate for most data centers. The challenge is making sure each VM is getting the right level of performance at the right time. Again there are a group of VMs that simply will never need flash performance, so they can be excluded from the conversation, put them on capacity storage and move on.

Managing Performance Sensitive VMs

There is also a group of VMs that, while not always needing flash performance, should be placed on flash to make sure that there is no risk of a user noticeable performance drop that impacts productivity. For these VMs, AFAs are appropriate but a hybrid system, one with flash and HDD, could provide a similar assurance by locking these VMs in flash. Using a hybrid array with flash pinning is ideal if there are just a handful of these performance sensitive VMs in the data center.

Troublemakers

Then there is a larger segment of VMs that occasionally need high performance and these VMs are the trouble makers. If an AFA is selected, these VMs will go on flash even though they really only need flash performance occasionally.

If a hybrid array is selected these VMs will reside on HDD most of the time and then be promoted to flash when their access activity increases. The users will experience a performance difference depending on how long it takes that VM’s data to be promoted to the flash storage area. Remember that for a hybrid system, the promotion is within the same system and the entire VM does not need to be promoted to flash, just the parts of it that are being frequently accessed. In other words, this promotion should be able to occur relatively quickly.

For many VMs this performance variance during promotion, while noticeable, happens quickly enough that it does not negatively impact productivity. For the few VMs where the variance in performance during promotion time is a productivity issue, these VMs will also then need to be locked into flash or pre-promoted to flash if the workload spike can be predicted.

A key to managing through the troublemakers is a robust reporting tool that can provide insight into these VMs as to when performance spikes occur and for how long. This allows storage administrators to predict which VMs should be locked into flash or allow them to pre-position the VM in flash prior to a spike occurring.

StorageSwiss Take

All-Flash Arrays do make life simpler by addressing the performance problem permanently. If you can find an all-flash array that meets your capacity demands within your budget parameters then it is the way to go. The simplification alone is worth it. If your budget won’t allow the AFA choice then look at a hybrid array, but still buy as much flash as you can afford. Doing so will make it easier to lock performance sensitive workloads into flash.

Wait! There are more flash and virtualization myths to be exposed:

• Myth or Fact? Flash storage is all the same
• Myth or Fact? Deduplication is all the same
• Myth or Fact? Integration is all the same

Watch the on demand webinar to see as we expose these myths.

About George Crump

George Crump is the Chief Marketing Officer at VergeIO, the leader in Ultraconverged Infrastructure. Prior to VergeIO he was Chief Product Strategist at StorONE. Before assuming roles with innovative technology vendors, George spent almost 14 years as the founder and lead analyst at Storage Switzerland. In his spare time, he continues to write blogs on Storage Switzerland to educate IT professionals on all aspects of data center storage. He is the primary contributor to Storage Switzerland and is a heavily sought-after public speaker. With over 30 years of experience designing storage solutions for data centers across the US, he has seen the birth of such technologies as RAID, NAS, SAN, Virtualization, Cloud, and Enterprise Flash. Before founding Storage Switzerland, he was CTO at one of the nation's largest storage integrators, where he was in charge of technology testing, integration, and product selection.