As the number of cores per processor coupled with the amount of memory on those processors continue to increase, it’s only a matter of time before developers routinely make greater use of in-memory computing. The simple fact of the matter is that the fewer times an application needs to make calls to disk-based storage system, the faster it runs.
To help IT organizations take better advantage of this fundamental shift in enterprise computing, VMware today released version 7.0 of the VMware vFabric GemFire in-memory data grid, which can now support up to 40TB of data running in memory. While in-memory computing was once seen as an expensive adjunct to the database, Dave McJannet, VMware director of cloud and application services, says in-memory computing is rapidly becoming the data store of record in the enterprise.
This has significant implications for how databases are constructed, deployed and managed. Increasingly, more of the advanced data management functions that used to be handled by a database running on top of disk storage are being moved into memory alongside the application. In effect, the role of the traditional database is being reduced to managing data at rest in storage, while all live data is processed in memory, or as McJannet puts it, “Data lives in memory and sleeps on disk.”
Applications that take advantage of in-memory computing will typically run 40 times faster than they do on disk. In addition, that kind of processing capability makes it possible to run, for example, transaction processing and analytics code in parallel within the same application, thereby being able to apply analytics in real time. Essentially, says McJannet, we’re witnessing the emergence of fast data at the same time as Big Data. In most instances, Big Data will be processed on a Hadoop cluster, but there are going to be instances where Big Data and fast data are also going to be currently processed in memory.
McJannet says that unlike in-memory computing approaches that are tied to a specific hardware platform, VMware vFabric GemFire assumes that the underlying IT environment is based on traditional x86 servers deployed in a scale-out architecture. As such, VMware vFabric GemFire is going to be much more broadly applicable, says McJannet.
New data grid frameworks such as VMware vFabric GemFire are on the cusp of fundamentally transforming how data is deployed and managed across the enterprise. The question at this point is not whether enterprise IT organizations will be making the move to some form a data grid running in memory, but rather when and to what degree.