Amazon Web Services option gives easier entry to SAP's HANA

An early adopter of HANA One on AWS reports success with the service

By , IDG News Service |  Software

SAP's HANA in-memory database has become the centerpiece of the software vendor's technology strategy, both for high-speed analytics and in support of transactional applications like its flagship Business Suite. But HANA, when sold in its appliance form, can fetch a steep price tag.

In October, SAP made a limited version of HANA available on Amazon Web Services. Dubbed HANA One, the service limits database sizes to 60.5GB and SAP recommends customers stick to 30GB or less to ensure top performance. But customers are also allowed to run applications in development, testing and production environments, and can pay for the service by the hour instead of forking over a large up-front chunk of money. Prices start at US$0.99 per hour, plus Amazon's infrastructure fees.

An early adopter of HANA One is the Canadian newspaper the Globe and Mail.

The paper had attempted to conduct deeper analysis on clickstream data captured on its website by its Omniture software, but had difficulty pushing the multi-gigabyte daily data loads of unstructured information into its Oracle database, said Sandy Yang, SAP Business Warehouse functional analyst at the paper, during a webcast event this week.

The Oracle database has trouble with the "schema-less" information output from Omniture, said Kevin Schlueter, enterprise architect at the newspaper. "Oracle isn't quite the right tool for this job. It's a great tool for a lot of other things, but not necessarily this."

Omniture provides a Web interface that can show metrics such as page views per second and the number of unique visitors to the site, but in many cases, the Globe and Mail's business analysts have deeper questions that require further analysis on the raw data, Yang said.

Ultimately it was decided that the paper would build out data analysis environment on Amazon Web Services, using its Elastic MapReduce service, which incorporates the open-source Hadoop framework for large-scale information processing.

But Hadoop is a batch-processing framework, making it a challenge for analysts to post "what if" type questions about the data, Yang added. SAP, meanwhile, has held up real-time data analysis of that nature as one of HANA's strong suits.

Yang learned of HANA One's availability on AWS and started up an instance. Within several days she had managed to import one day's worth of Omniture data into the instance and create a first analytic view, she said.

Yang was then able to sit with analysts and give them answers to their questions about the data quickly, rather than having to go back and run additional EMR jobs based on what they wanted to know.

Join us:






Answers - Powered by ITworld

Ask a Question