Saturday 13 March 2010
In-memory data grids store information that applications need in memory across a pool of servers, instead of reading it off disks, resulting in major performance gains. The Coherence product is one of the more mature in a space occupied by offerings from IBM as well as smaller companies like GigaSpaces and a number of open-source projects. Microsoft is also developing a system dubbed “Velocity.” o date, such systems have breathed rarefied air, mostly supporting large-scale Web sites and high-throughput transactional systems, such as stock trading applications. The in-memory data grid market in total generated less than US$100 million in revenue during 2008, according to Gartner, which prefers the term “distributed caching platforms.
But some observers believe that in-memory approaches to data management could eventually gain serious traction in cloud-computing deployments. The technologies would help applications that weren’t originally designed for elastically scalable infrastructure systems like Amazon Web Services to run more effectively, albeit “via some re-engineering,” Gartner analyst Massimo Pezzini said in a recent report. Beyond vendor trepidation, other factors stand in the way of broader customer adoption, such as the complexity of deploying and managing the systems and limited support from systems integrators and ISVs, according to Pezzini.
