There has been more news in the recent weeks on SSDs, or more specifically, the entry of more memory-based solutions into the data center. Intel announced some of their plans as part of their developer conference, IDC has firmly created a category to watch, research, and forecast, and IBM announced Project Quicksilver...no doubt their way of capturing some of the industry buzz.
Memory in all its shapes and forms is advancing in the datacenter, displacing other media types that are no longer effective in getting the job done. This is a naturally occurring technology cycle that has and will continue well beyond this hype-fueled flash frenzy. But that doesn't stop industry players from promoting and positioning their solutions in ways that appeal to specific audiences. MLC flash for this SLC flash for that, battery backed DRAM for this, phase change memory for that, and the list goes on.
On the placement side, there is a race to own as much of the newfangled media as possible. Here, I'd like to comment on a few inevitable truths:
- If you make your livelihood selling servers, then servers are the best place to put all this new memory
- If you make your livelihood selling disk arrays, then disk drives and arrays are the best place to put all of this new memory
- And if you aim to add new value to the data center in ways that break away and improve upon conventional architectures, then you put it somewhere in between.
End users will have to make up their own mind. We have written extensively on the pros and cons of the various approaches, all of which are valid, but serve different markets, segments, and customers. For a more detailed discussion and perspective, please see Memory in the Data Center - The Series.