What Solid State Form Factor Is Best - Integration

Returning to our Solid State Form factor series; this entry we are going to begin the discussion about solid state integration. There are really two parts of the integration discussion; how will you integrate solid state disk into your storage infrastructure and the other is how will your vendor integrate solid state disk into their storage system? We'll tackle the vendor issue first since it may directly impact how you integrate solid state storage.As we discussed in our recent webinar "The State of Solid State Storage" how the vendor decides to integrate solid state disk into their platform is going to impact performance and cost. Here specifically I mean how legacy storage vendors integrate solid state into their storage platforms, not solid state systems or memory arrays. Legacy systems have the bulk of the integration work to do since they are typically using the disk form factor version of solid state known as SSD.

Much to the apparent surprise of many storage suppliers inserting SSD into a legacy storage system is not the full extent of the integration effort. Just because it fits in the drive slot does not mean your integration work is done. There is more to it than that. First there are the performance issues. In most cases it only takes a small number of SSDs to be able to produce more I/O performance than the typical storage controller can support. Systems that are going to support SSD for anything other than cache should be able to have multiple controllers and I/O paths to the storage enclosures.

There is also the issue of the storage enclosure. In most cases you can't populate the whole shelf with SSD. The combined I/O of the drives is greater than what two or four I/O ports coming out of the shelf can deliver. We have actually seen configurations from legacy storage vendors with only 2-3 SSD drives in a shelf and the rest of the shelf empty, violating every law of efficient space and power utilization, not to mention the cost of having to pay full price for half used shelves.

This is not necessarily the fault of the SSD manufacturers themselves. The drives are doing exactly what they said they would do. The challenge is on the legacy storage vendors to design enclosures that are tuned to the capability of the drive technology. This means either faster shelves, less expensive shelves that support less drives and require less power or use solid state for cache only.

These are just a few performance issues that legacy storage vendors need to design around. In a future entry we will discuss the challenges to applying RAID type of protection in legacy storage systems when using SSD.

For now what it comes down to is that to take full advantage of SSD, a legacy storage system is going to require some kind of scale out storage, at a minimum at the controller and quite possibly on the storage enclosure. There are alternatives, you can by a solid state only appliance to either address a point specific problem, a cache like front end that can provide broader performance improvement while keeping your legacy storage or a solid state only system that has the full compliment of storage services (snapshots, replication). Each have their fit and we will cover them in upcoming entries as well as how to integrate them into your enterprise.

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.