Why AMD Dumped Oracle for Hadoop
Courtesy Jack Clark,theRegister.co.uk
AMD has migrated terabytes of information from an Oracle Database installation to an Apache Hadoop stack, claiming Oracle’s pricey software was suffering from scaling issues.
The chip maker’s chief information officer, Jake Dominguez, revealed further details of the transfer in a chat with The Reg.
"Within the common Oracle platform we had we were struggling from a performance and reliability perspective," Dominguez told your correspondent in Atlanta just before the weekend. "One of the areas we were struggling with was in our test and assembly manufacturing – large, large datasets."
The migration of 276TB of data, which was completed last year, was prompted by "an environment outage that took weeks to recover," according to an internal document seen by El Reg. This encouraged AMD to replace Oracle for something else.
In the end, the processor giant settled on using Cloudera’s Hadoop distribution along additional open-source projects Apache Hive, ZooKeeper, HBase, HDFS, httpfs, LZO compression, MapReduce and others.
According to AMD, the Hadoop software has an unlimited row limit for query results compared to 100,000 rows on the chip giant’s Oracle setup, and "99 per cent of all queries execute in 15 minutes or less, with a median execution time of just 23 seconds."
What makes this shift so significant is that Oracle wants you to think AMD is the sort of company that will always use Oracle kit.
Oracle is grappling with a shift in the data warehouse and analytics market: its core business is being squeezed by free and open-source on-premises software, and its cloud wing is facing off with Amazon Web Services and the like.
Many organizations have sought to extricate themselves from Oracle’s grip, either by swapping out Oracle-owned open-source tech for other software, as Google did with a vast MySQL to MariaDB migration, or by shifting away from the company’s proprietary databases to open-source ones, as the UK’s National Health Service did with a major Riak migration.
One of the main open-source technologies commonly being deployed to supplement or replace Oracle is Hadoop, a data storage and processing framework that was first developed at Yahoo! in 2005 by engineers attempting to replicate some advanced technologies invented at Google.
Today, software like Hadoop, and other distributed data storage and management frameworks like Cassandra and Riak, are competing with software from IBM, SAP, and most prominently Oracle.
For AMD, a sophisticated multinational manufacturing company, to launch a major Oracle migration project is representative of a broader shift in IT which benefits low-cost or free software at the expense of incumbents like Oracle.
"We made the pivot to Hadoop [and] it not only increased our reliability but [improved] our response time," Dominguez told us. "It’s going to be an integral part of our enterprise data warehouse concept." ®