Amazon's Relational Database Service Adds High-Memory Instance With 244GB Memory

Started by TechCrunch, Sep 05, 2013, 02:32 AM

TechCrunch

Amazon's Relational Database Service (RDS), which gives developers access to MySQL, Oracle or Microsoft SQL Server database engines, added a new instance type for applications that need a lot of memory. With 244GB of memory, the new High-Memory Cluster DB Instance Class offers 3.6x as much memory as the "quadruple extra large" instance, which has 68GB of memory. But for now, this new instance only supports MySQL databases.

The new instance is also significantly more powerful and offers — using Amazon's measurements — the equivalent compute power of 88 2007 Intel Opteron or Xeon processors running at 1.0-1.2 GHz. Amazon says it uses two Intel Xeon E5-2670 eight-core processors to power these instances, which operate on a high-bandwidth network. Overall, the company argues, this new instance will allow developers to realize speed increases of about 60 percent compared to the smaller instance. As Amazon's Jeff Barr notes, he believes that this new instance type "opens the door to database-driven applications that are even more demanding than before."

As Amazon notes, this massive amount of memory will allow developers to configure MySQL to fit most online transaction processing databases into memory, which should make this instance type perfect for this kind of usage. Because Amazon's RDS offers support for SQL and Memcached API access, developers can use the service for complex SQL transactions, as well as high-performance NoSQL operations that draw from a single database.

Unsurprisingly, this new instance type also comes at a relatively hefty price, starting at $2.074 per hour with a three-year commitment and just over $3,000 in upfront costs. On-demand usage in Amazon's US East and West regions start at $4.725 per hour.



TechCrunch