Why opt for SSD in the enterprise?

Tony Afshary gives his reasons for, and rules on, effective SSD deployment

As the cost of flash comes down, more SSDs are being deployed in a tier-one space, where applications require high random IOPS performance to handle intense transactional processes that generate revenue.

PCI Express-based storage adapters with integrated solid-state modules boost storage performance the most for the smallest storage footprint, and SSD is a viable alternative to HDD.

Solid-state storage can be implemented as a dedicated storage volume in the same way as traditional rotating media. It is comparatively expensive, but the performance gains and latency reductions are dramatic, particularly in application environments where data needs to be processed and analysed in real time. It also saves on power, cooling and maintenance.

Together with RAID controllers, particularly those with dual-core processors, SSD can provide amazing performance for web, file, SQL and other transactional applications. If the RAID controllers are coupled with SSD optimisation software, you can have high IOPS and low latency.

Intelligent tiering software can move data between the various storage media volumes dynamically, making up a total pool of storage where users can store the most-accessed data on the highest-performing media, with the rest allocated to cheaper volumes.

When access to all application data is mission-critical or the highest possible performance is required, all application data should be stored on a solid-state device.

Solid-state cache memory can allow the caching system to observe the data access patterns and determine what data to store where for maximum performance. The only task for the customer is to determine and configure the caching system to use the desired amount of SSD.

In my view, 6Gbit/s is best. SSD delivers the best user experience if paired with 6Gbit/s RAID technology.

However, solid state – like HDD – can still fail and will eventually wear out due to write endurance limitations. The data still needs to be protected and monitored, using high-availability RAID algorithms and software features that copy data automatically from an SSD that has developed reliability issues to a spare or new drive.

A predictive failure event notification or SMART command can initiate a rebuild automatically, preserving the data on a failing or underperforming SSD.

Tony Afshary is marketing director for accelerated solutions at LSI