Most IT organizations operate under the assumption that achieving maximum application performance requires a relational database. But increasingly, a handful of NoSQL database alternatives have shown that they can more efficiently meet the most pressing application performance demands.
One of those offerings comes from Aerospike, which today unveiled an upgrade to its namesake NoSQL database that runs in-memory. Version 3.5 of Aerospike adds a disk write cache that allows reads to be done from recently written data and a new memory allocation system that greatly reduces DRAM fragmentation.
In addition, Aerospike offers integration with Hadoop, has enhanced the clients it supports and has simplified the installation process using options such as Docker containers or Vagrant images to deploy Aerospike.
Aerospike CTO Brian Bulkowski says that IT organizations have the option to deploy the Aerospike open source database on premise or in the cloud to support applications written in a wide variety of programming languages. Bulkowski says that Aerospike doesn’t have a vested interest in forcing that issue one way another, but concedes that the forces of data gravity are conspiring in a way that will continue to push more data into the cloud. With more applications running in the cloud, it’s only a matter of time before more of the applications running on premise move to the cloud to simplify data integration, notes Bulkowski.
The most unique thing about Aerospike, says Bulkowski, is that it can run as many as 2.5 million transactions per second (95 percent reads/5 percent writes) using just one bare-metal x86 server. For IT organizations, that means they need to deploy fewer servers within their application environment than they would running other types of relational or NoSQL databases, says Bulkowski.
When it comes to databases, there is a lot more diversity than ever. But when it comes time to scale an application, the underlying database clearly makes a difference in terms of the performance of the applications and the total cost of the server environment that needs to be constructed to make that performance possible.