MemSQL Pushes Translytical Database into the Cloud

MemSQL today unfurled a preview of a cloud service based on a database optimized for translytical applications.

The MemSQL Helios service is based on the beta edition of a forthcoming 7.0 release of the MemSQL database, a distributed relational database optimized to run operational analytics applications in memory.

See also: Infrastructure Architecture Requirements for Continuous Intelligence

The latest version of MemSQL also provides for the first time a “SingleStore” capability that eliminates the need to choose between a row store or a column store for different classes of workloads. That capability should reduce a lot of the performance tradeoffs organizations now make when building and deploying translytical applications, says Peter Guagenti, chief marketing officer for MemSQL.

“There used to be two distinct stores that needed to be managed,” says Guagenti.

Version 7.0 of MemSQL will also bring support for incremental backup and synchronous replication capabilities with virtually no performance penalty, adds Guagenti. Those capabilities are expected to increase the appeal of MemSQL for use within Tier One application environments that tend to be systems of record within the enterprise, says Guagenti.

MemSQL Helios, meanwhile, is a database service that can be hosted on a public cloud and managed on behalf of customers by MemSQL. That approach eliminates the need for an organization to hire a dedicated IT team to manage the infrastructure on which the database runs, says Guagenti.

In many cases, translytical databases also eliminate the overhead associated with extract, transform and load (ETL) processes that are usually relied on to move data from a relational database to a data warehouse platform that is typically based on a columnar database.

The cloud service being provided by MemSQL on an invitation basis is intended to target a massive wave of applications that are now being deployed in the cloud. While most relational database applications deployed today still run in on-premises IT environments, MemSQL is trying to attract emerging cloud-native applications to its platform that require analytics to run in real-time as transactions are being processed. MemSQL chose a relational database as the foundation of that platform to ensure the transactions being processed were full ACID (atomicity, consistency, isolation and durability) compliant. MemSQL claims it can already process one trillion rows of data per second.

It’s still not quite clear what percentage of applications will require translytical capability. Many proponents of these databases argue that the rise of digital business processes makes their adoption all but inevitable. That shift to a new database platform capable of processing massive amounts of data is also something of a symbiotic driver for adopting public clouds, which make it less expensive to store all that data than an on-premises environment.

Being able to scale these applications in memory on x86 processors is only a relatively recent phenomenon. Translytical applications have been running on mainframes for some time now, but the cost of those platforms has tended to limit adoption.

The goal now is to not only reduce costs by eliminating the need to deploy and manage separate databases, but also build new classes of applications that given the costs involved would not previously have been thought feasible.