Oracle (NASDAQ: ORCL) is updating its Coherence in-memory database this week, providing enhanced scalability.
Oracle Coherence 3.7 includes new Elastic Data features that enable enterprises to maximize memory and solid state disk for performance.
"There are about 500 improvements in the Coherence 3.7 release, so it's a pretty substantial release," Cameron Purdy, vice-president of development at Oracle, told InternetNews.com. "Headlining the release is a feature we call Elastic Data, which is the culmination of a number of years of development."
Purdy said Oracle Coherence customers are managing lots of different types of information and for each type they have a set of information. Previously for each type, they had to define how much memory they needed on a cache-by-cache basis.
With Elastic Data, when a server comes up, it will now be able to allocate a specified amount of memory. Whatever it cannot accommodate now goes into flash storage.
"So this is flash Solid State Disk storage that transparently manages information across a cluster and between memory and flash," Purdy said. "We're using Java NIO as the basis of this implementation and we're able to do write to disk at full disk speed."
According to Purdy, on Oracle's own benchmark testing there is virtually no performance difference by using the flash storage in combination with memory.
"What it means for customers is that instead of configuring things on a set of data basis, the system automatically manages and prioritizes the memory, and it's all done transparently to the application," Purdy said. "So with Elastic Data, it's dramatically simpler to configure and manage and we can now also support dramatically larger amounts of data, terabytes of data in memory."
Coherence 3.7 and its Elastic Data capabilities have been optimized to run on Oracle's Exalogic servers, though Purdy noted that other hardware is also supported.
Purdy added that Coherence is using clean algorithms to mitigate the risk of SSD failures and to reduce the wear on the drives so that they last longer.
Oracle is also making Coherence more attractive to developers with its enhanced Extend client libraries. Purdy explained that Extend allows applications written in multiple languages to access Coherence.
"We've also added dynamic load balancing," Purdy said. "We can support thousand of servers with Coherence and we can also support thousands of clients connecting into any of those servers."
Purdy added that in Coherence 3.7, clients can take advantage of WKA (Well Known Addresses), which is a simple way to do discovery within a large environment. As clients connect in, Coherence load balances based on demand and availability. Oracle has also done integration with third-party load balancing hardware, including F5's Big-IP platform.
Moving forward, Purdy noted that Oracle is looking to do more than just manage information in memory with Coherence. He noted that Oracle is always looking for ways to make it simpler for developers to more easily put together information and access the data from Coherence.
"The goal of all this, is to have a simple to manage and monitor product that supports developers," Purdy said. "Trust me there is plenty of work left, we're not going to get bored."
For more on in-memory databases and analytics, see Analytics and In-Memory Databases Are Changing Data Centers