Oracle, NoSQL, and 40TB Tables
With Oracle’s fourth top line miss in nine quarters just now behind us, financial and IT analysts are providing interesting insight into what they think the causes are behind the financial shortfalls of the biggest database vendor on the planet. In his report on Oracle’s recent earnings announcement, Peter Goldmacher says this:
“We believe that we are in a major technology transition as IT buyers are increasingly turning to newer apps and data mgmt technologies that offer more robust and flexible functionality at dramatically lower prices.”
Here at DataStax, I can tell you that’s exactly what we’re seeing.
If you’re thinking, “What else would a NoSQL vendor say?”, that’s OK because I’d be thinking the same thing if I wasn’t living it everyday. But facts are facts.
We received an overwhelming response to our recent announcement of major companies switching from Oracle to NoSQL, and specifically Apache Cassandra and DataStax Enterprise. Stories like those that ran in USA Today detailed how modern enterprises such as Netflix, Ooyala, and Openwave have moved from Oracle to DataStax for both technical and cost reasons.
On the technical side, companies are finding that RDBMS’s like Oracle run out of steam when taxed by their big data, line-of-business applications. Although the technical reasons for making a switch to NoSQL vary depending on the use case, the pattern we see again and again include the need for:
- Continuous availability with zero tolerance for downtime.
- Handling extreme amounts of incoming, high velocity data.
- Supporting “Active Everywhere”, which means writing and reading data in many different locations at once.
- Spanning multiple data centers and/or cloud availability zones.
- Keeping large volumes of data online at all times.
- Supporting all types of data in a more fluid and flexible data model.
- Scaling in ways that “future proofs” applications.
We hear these requirements come up constantly in the customer and user interviews we do with companies that are using Oracle, DB2, MySQL, and other RDBMS’s. Just with MySQL, you can read story after story from companies like Boxever, Bazaarvoice, Easou, Healthcare Anytime, , Pantheon, Metabroadcast and others on how and why today’s changing data landscape necessitated a move to Cassandra and DataStax.
The great news is, not only are companies able to now meet the technical requirements of their modern line-of-business applications with DataStax, they’re saving a considerable amount of money in the process. Just one example is Constant Contact who swapped out DB2 for DataStax Enterprise and reduced their database costs by 90% over what they were paying IBM.
Walking the Big Data Talk
When considering a move from Oracle, you need to be careful that you’re not trying to shove a square peg into a round hole and ensure the technology you’re using is really up to the challenge of both meeting the requirements I listed above and is actually ready for enterprise database battles. So many times, software vendors will make claims that they were born for big data, when in reality they have many of the same foundational architectural limitations as legacy RDBMS’s (e.g. master/slave architectures, the need for manual app sharding, etc.)
Make no mistake, while replacing smallish Oracle databases with other solutions may work, going after large, enterprise-class applications that have been backended with Oracle is another story. For those, you’ll need an industrial-strength NoSQL solution like Cassandra that has actually been architectured from the ground up to handle today’s big data workloads.
A case in point is eBay. The company swapped out Oracle with DataStax for a number of application use cases, and is currently managing hundreds of terabytes on the DataStax Enterprise platform.
During their recent Cassandra Summit talk, eBay described how they have one routinely accessed table in a Cassandra database that’s 40TB in size. No, that’s no misprint – one table that’s 40TB.
Now understand that handling big data means more than just taking on large data volumes – it also involves having other capabilities like efficiently dealing with high velocity data, writing data all over the globe and synchronizing it, storing all varieties of data, and more. And Cassandra does all that…plus, it allows you to enjoy fast performance and linear scale to the point where you can have 40TB data objects in your line-of-business applications that are accessed as matter-of-factly as objects much smaller.
In short, Cassandra walks the big data talk and does so everyday.
How to Get Started
So if you’re considering a move from Oracle or any other RDBMS, what should you do? First, educate yourself so you’ll know if your situation will benefit from using NoSQL over Oracle. For that I recommend viewing our webinar “How to Tell if Your Business Needs NoSQL” and reading our “Why NoSQL?” white paper.
Next, become familiar with how NoSQL data models differ from an RDBMS. One of the keys to success in NoSQL is understanding the data model and getting your design right. For that, I recommend viewing our webinars “The Data Model is Dead, Long Live the Data Model” and “Become a Super Modeler”.
Finally, download DataStax Enterprise, review our online documentation, and give NoSQL a try in your shop or in the cloud. You can learn about and use DataStax Enterprise in your development environment with no charge whatsoever.
Once you’ve done these things, I think you’ll agree with Peter Goldmacher that a technical transition to modern technologies like DataStax Enterprise and Cassandra is the reason why Oracle’s taking it on the chin these days.