What Solid State Storage Form Factor Is Best? Part II

As discussed in an earlier entry, there are three basic types of solid state form factors available in the market today; PCIe as we discussed last entry, Solid State Disks, which we will cover in this entry and Solid State Appliances also called Memory Arrays which we will cover next. We'll conclude this series with a discussion in integration methods that storage vendors are using to implement solid state technology.Solid State Disk (SSD) is the term often used to blanket the whole solid state storage market. Really it should be isolated to the devices that are in a hard drive form factor and are designed to be installed where hard drives typically installed. Instead of platters that spin they are have memory with no moving parts. For obvious reasons these types of technologies appeal to both users and system manufacturers. Simply plug in the SSD where you used to have a hard drive and you are off to the races.

The SSD form factor is gaining in popularity in both the laptop market as well as the server market. As we discuss in our article "Improving Storage Performance in the Medium-Sized Enterprise" these can be ideal solutions for mid-sized data centers looking to solve specific server based challenges like swap-file performance and transaction logs. These environments that are have yet to implement shared storage can benefit by using SSD as a caching drive against a larger internal RAID array or as a drive to store swap files. In both cases there is limited need for redundancy since the SSD would only be storing temporal data.

If there is a shared storage system in place, and many smaller data centers get there sooner than ever thanks to server virtualization, SSDs can still play an important role. While this smaller SAN can't justify SSD in the shared storage system itself, installing them in the attaching servers may be a perfect compliment, especially in 1GbE iSCSI shared storage environment where network performance is fine for most workloads but can occasionally get overrun.

In the past most small SANs kept boot files and swap files local while the more capacity centric data is on the shared storage system. Moving to local SSD in these attaching servers allows for faster boot times and keeps the heavy I/O traffic of swap files off of the 1GBE iSCSI segment. This clears the network for the rest of the storage I/O. The local capacity needed on the server as a result is relatively small. Instead of buying a mirrored set of mechanical drives and wasting that capacity, buy a set of smaller but significantly faster SSDs to not only improve performance of the server but improve the performance and scalability of the SAN. With mirroring in place the SSDs can be used for data that is a less temporal like transaction logs that databases and email systems create, further reducing I/O traffic to the shared storage system.

Finally SSDs are also used as a gateway to solid state technology in traditional enterprise storage systems. While this provides a fast track to solid state it is not without limitations. We will discuss the pros and cons of using SSD in traditional storage arrays in an upcoming entry on integration methods.

SSD technology, especially in servers, to either improve the performance of local mechanical arrays via caching and storage swap files or to extend the scalability of a small SAN by offloading storage traffic brings immediate value and is something the mid-sized data center should definitely explore.

Track us on Twitter: http://twitter.com/storageswiss

Subscribe to our RSS feed.

George Crump is lead analyst of Storage Switzerland, an IT analyst firm focused on the storage and virtualization segments. Find Storage Switzerland's disclosure statement here.