Memory in the computing clouds?

Time for a technology posting! A dearth of meetings on my calendar today gave me a chance to catch up, and then to go and read some of my favorite RSS feeds and troll for some useful perspectives on where computing is heading. I’m more and more convinced that the SaaS (software as a service) industry is going to make it this time (unlike the first Internet boom of the late 1990’s). Physical infrastructure is more robust, connectivity is better, and application architectures are much more sophisticated and capable. One thing that caught my eye today was a cloud computing blog posting that referenced a great article on Cloud-Based Memory Architectures. I think that this is really where things have been headed for a while, we just didn’t realize it. Need performance? Big disk arrays with more cache. Hold indexes in memory. Are you I/O bound in your app? Odds are, yes…I’ve seen that time and time again, even when I thought from looking at the system performance tools that I was not I/O bound. Anyway, the article referenced above is a great set of thoughts and links on this subject. It’s a long read, but if you are interested in application architectures, it’s worth it.

How do I see this trend affecting computing at organizations like mine, UNCG? I think that we’ll see more applications moved to the cloud, not just for cost savings in hardware and support staff, but for performance. With big pipes we can consolidate in ways that leverage business/organizational scalability, and by moving to applications where the canonical copy of the data is in memory, not disk, we’ll gain substantial performance benefits. We need to be prepared to continue to move applications away from our own machine rooms. As this develops, though, it will be interested to see how issues surrounding security and backup will be handled…

Leave a Reply

Your email address will not be published.