Oracle on Monday announced an in-memory option for Oracle Database 12c that will accelerate analytics, data warehousing, reporting and online transaction processing for Oracle in-memory applications.
These applications “deliver extreme performance on Oracle Engineered Systems,” the company said.
“This is a story about Oracle being a bit late to the game in incorporating in-memory into its database,” remarked Wayne Kernochan, president of Infostructure Associates.
Oracle has had in-memory technology since it acquired TimesTen in 2005, but “what it hasn’t had is in-memory integrated into Oracle Database,” Kernochan noted. “Still, this is a good thing for their customers — assuming that unlike Oracle’s OLAP Option in the past, the price isn’t too steep.”
With this development, “Oracle Database itself acts as an in-memory RDBMS, with both row and column internal formats,” Carl Olofson, a research vice president at IDC, told TechNewsWorld.
“What this really means is that Oracle Database users get the benefits of [an in-memory database] without changing their SQL,” he continued.
The announcement “is an opportunity for Oracle to sell platforms and not just software,” opined Jim McGregor, principal analyst at Tirias Research.
What Oracle Said
Making the announcement at Oracle OpenWorld 2013, under way in San Francisco through Thursday, Oracle said that every application that runs on Oracle Database 12c can automatically and transparently take advantage of the Oracle Database In-Memory option.
Existing applications will retain full database functionality while speeding up. New applications that were previously not practical because of performance limitations will be developed.
Oracle’s JD Edwards EnterpriseOne In-Memory Project Portfolio Management and In-Memory Sales Advisor, Oracle’s PeopleSoft In-Memory Project Discovery; Oracle’s Cost Impact Simulator and Gross Profit Analyzer, and the company’s Cost Comparison Tool are some of the applications that will benefit from the Oracle Database In-Memory Option.
Oracle’s announcement “is just an evolutionary change, not revolutionary,” McGregor told TechNewsWorld. “It’s a change you should expect. With each generation of the software, it has to get better.”
What’s an In-Memory Database?
In-memory databases, or IMDBs, are also known as main memory database systems (MMDBs) or memory-resident databases.
They store computer data in main memory rather than on hard drives. This makes them faster than disk-optimized databases, because “in situations where only main memory is needed, [IMDBs] can assume a flat memory without a need for all sorts of code to handle uploading and downloading data from and to disks,” Kernochan told TechNewsWorld. “That code wastes a lot of time.”
By eliminating the code, IMDBs “can be one to two orders of magnitude faster” than disk memory, he pointed out.
“Practically speaking, vendors are now talking about main memory plus solid state disks being able to handle 80 to 90 percent of apps,” Kernochan continued.
The Blots on IMDB Technology
One problem with IMDBs is that they store data on volatile memory devices, so they lose all stored information when the devices power down or are reset.
Another is that main memory costs are high.
However, main memory prices are decreasing faster than disk prices are, Kernochan said.
The speed advantage IMDBs now enjoy over relational database management systems (RDBMSs) may not last.
“It is a general truism that about 80 percent of enterprise data warehouse data is very infrequently accessed, and so, with proper management, a memory-optimized RDBMS can keep most of its data on disk and still deliver performance benefits that are nearly identical to pure IMDBs,” IDC’s Olofson pointed out.
IMDBs and Big Data
IMDB technology will only have a minor impact on Big Data applications, “simply because SAP and IBM, with BLU Acceleration, already have in-memory technology in-database,” Kernochan suggested.
How much IMDBs will impact Big Data “depends on the application and how it’s targeted towards that,” Tirias’ McGregor said.
“Oracle, being the database company, they’re going to get better and better at this so they can analyze the data you’re using and minimize the number of hits to your hard drive by moving the data with the most data against it to memory.”