Sign in with
Sign up | Sign in

Why You Need Enterprise-Class

Enterprise SSDs: How to Prove Real World Performance
By
Brought to you by What's this

Consistency in storage devices is paramount within enterprise settings, but one should also question why enterprise drives perform as they do. Obviously, enterprise drives cost more, but an increasingly number of SSD vendors are taking client drives, tweaking them, and marketing them to enterprises. Are there physical or component-level differences between the two.

The answer is a clear yes, and the details begin with silicon choices. SSD controllers are not a one-size-fits-all affair. As with most things, ASIC manufacturers create different model grades according to computing bandwidth, algorithm sophistication, and other factors. Enterprise-class drives almost invariably use the highest grade of controllers available.

“Another difference between client and enterprise SSDs is that a lot of the enterprise drives either have non-volatile cache or they have some sort of power-protected write cache,” says Seagate’s Jeff Nowitzke. “Those are extra expenses that normally aren’t going to end up in your client products.”

These cache types enable enterprise-class SSDs to perform in what is functionally a write cache-enabled mode, which assists with drive performance. However, this mode often leaves data exposed to a risk of loss in the event of power failure. Prior to enterprise SSDs with power loss protection, most business customers would run drives with write cache off for maximum data integrity. Meanwhile, client devices are typically configured with write cache enabled in order to achieve peak performance. The risk in this case is that an unexpected power loss will destroy any data sitting in cache. Enterprise vendors are now taking care to provide some method of power to enable write cache such that data will be saved securely in the event of a sudden power cut.

The amount of free space maintained on an SSD directly affects its endurance, or expected number of sustained write cycles until NAND failure. Consumer drives often come from the factory with only 7% overprovisioning, which refers to the amount of capacity that users cannot address but that stays invisibly available in the background for drives to use in streamlining writes. More overprovisioning means longer endurance. This is why enterprise drives often ship with overprovisioning of 20% or more. This will also help improve write performance.

These factors and others play into why enterprise drives specify more robust duty cycles. Whereas a consumer drive is likely made to run eight hours per day during the work week, an enterprise drive is expected to operate around the clock for years without rest. Moreover, the enterprise drive is built to run under this heavier load within a much warmer ambient environment. This is yet another element in the consistency argument that gets little attention. Enterprise drives are typically validated at 60˚C, not the 40˚C typical of client drives. Buyers should be asking if their storage devices can provide consistent performance under hot aisle conditions.

By now, it should be clear that there’s far more to an enterprise SSD than meets the eye—or even the spec sheet—and all of these factors feed into why an enterprise drive is able to maintain the sort of consistent performance reflected in SPC-1 testing. Before committing to a purchasing decision, talk with manufacturer representatives and find out how that vendor’s products fare with the criteria we’ve covered here and whether it has benchmark data that truly reflects your applications’ and environment’s needs.