IT is a global game of king of the mountain. The view is lovely from the top, but everyone else is looking to pull you down with new innovations in products, better services, and disruptive go-to-market strategies. Oracle has long held the crown for databases and a number of related business applications, yet is surely feeling some pressure from the ravenous hordes below. Two popular angles of attack have been in-memory and/or NoSQL databases. The goal of players such as Amazon (with DynamoDB), Apache Cassandra, MemSQL, MarkLogic, MongoDB, NuoDB, and many others has been to differentiate on the capabilities of their newer platforms to hopefully displace, or more likely find a niche alongside, the popular Oracle database. Even other industry titans like IBM (with DB2 BLU Acceleration) and Microsoft (with SQL Server 2014) have brought in-memory options to market with great fanfare.
Oracle has realized the threat and is now ready to defend its peak. With the public announcement of Oracle 12c this week, particularly the in-memory option, the playing field just got leveled. Or rather, remains tilted in Oracle’s favor once the GA comes later this summer.
In-memory is one of the hot areas in big data today. Everyone wants to capitalize on the increasingly affordable possibilities of cheaper DRAM. Particularly for analytics, being able to query columns of data without taking the expense of going to flash or (help us) traditional disk storage opens up massive performance gains. This enables you to look at a lot more data, a lot faster, getting to real-time insights even against unbelievable volumes of raw information. The business opportunities are myriad, many not even dreamt up as yet.
NoSQL brings the advantages in terms of web-scale, flexible schema-on-read, and combined real-time analytics even while loading new data from multiple sources and updating for live transaction processing. Devops shops love the stuff, but IT executives and operations admins aren’t always keen because of the relatively unproven reliability and limited management tools. It’s a maturity problem. Many of the newer technologies are rapidly addressing these gaps, but Oracle has had dozens of years experience in satisfying enterprise operational requirements.
So what do you get if Oracle can match your massive real-time performance gains by simply adding tables or partitions to in-memory column stores, drop the old indexes, and still retain all the availability and management functionality of the database? What if Oracle can do all of this plus run NoSQL and support it in enterprise requirements? What if Oracle can offer Cloudera Hadoop on the same purpose-built big data appliance? Or run 12c on a big memory appliance with 32 TB of RAM and 384 cores?!?
What if, indeed. Oracle can now bring the "real-time enterprise", and this will certainly help entrench and defend that coveted mountaintop position. The game plays on….