Financial institutions, marketing firms, and a host of others in the business of anticipating consumer patterns, depend on rapid analytic platforms to provide insights into their respective industries in lightening speed. The powerhouses behind this analysis were traditional databases, and among them, Oracle has been the leader. But that may be changing, in no small part to arch-rival SAP’s recent unveiling of HANA which allows complex and ad hoc queries of billions of records in seconds versus hours.
One of the chief problems with mainframes was that in storing application and transactional data in the same place, transactions and reports couldn’t be processed simultaneously, which created bottlenecks. Aside from purchasing additional mainframes to expand scale, companies developed the use of client servers to run one centralized database which could be grown by adding more servers. But even this didn’t completely solve for the problem in that relying on one database server eventually led to I/O stagnation, especially at OLTP (online transaction processing). As more and more users began to demand more complex and frequent reports, the problem only became exacerbated.
To address a trend of increasing internal and external pressures for data analysis and the inadequate capacity of I/O databases, SAP created HANA.
2011: The Year of HANA
When SAP announced this in-memory platform specifically designed for enterprise apps, chief among their claims was that it would turbo-charge the velocity of databases, as well as the front-end, providing the means to query billions of records in seconds. Add to that, it runs as a single database; no more data redundancy. Instead of one group of databases for transactional data and another for analytical data, both were run on the same database.
Oracle’s model was built on hyper data segmentation or specialization, e.g., one database for sales orders and another for product fulfillment. As a company grew, they would have to acquire more and more specialized databases – hence the data sprawl and crawl. Oracle has its own in-memory platforms Exalytics and TimesTen, but it is not throwing the cash cow of its original database platform model, which provides a stream of renewal income via licensing for each new database, out the window.
Some Hard Facts about Exalytics/TimesTen
The package appears to be more of an add-on, in that it still necessitates additional licensing for each segmented database, as well as the package. The duplication of transactional data between the application warehouse and Exalytics recreates the original problem of data redundancy (three sets of data and three levels of storage and servers). In comparison, HANA does not require redundant aggregates and indexes, or materialized views.
HANA vs TimesTen and Exalytics
- PR juggernauts aside, there are some clear cut differences between the two platform, such as:
- HANA manages data and accesses it in RAM, eliminating the need for MOLAP
- HANA can take on unstructured and structured data, parallel queries, whereas (for now) the TimesTen/ Exalytics package drops off at 1TB RAM (whereas HANA can store up to 500TB of high speed data).
- HANA’s price includes production and test/development requirements.
- And, HANA supports VMware vSphere.
- Archie Hendryx, “SAP HANA: A real-time challenge to the Oracle empire,” ZDNET, http://www.zdnet.com/sap-hana-a-real-time-challenge-to-the-oracle-empire-700000817.
- “On the Road to SAP HANA: Paving the Way,” Dolphin Corp., http://www.dolphin-corp.com/information-lifecycle-management/sap-hana.
About the author:
James Hadley is an IT consultant for many small to mid-size businesses. He lives in San Francisco with his wife and two sons.