SAP hasn't had a product as popular as its HANA in-memory data processing engine since the R/3 enterprise resource planning suite launched two decades ago and took off like wildfire, transforming the German software company into the dominant application software provider in the world. SAP wants to hit replay with HANA, and it is willing to shell out cash to make sure it happens.
The HANA in-memory database was developed by SAP and the project predates its $5.8bn acquisition of relational database maker Sybase in May 2010. The HANA database is a mix of various technologies, but significantly is based on an offshoot of the Adabas database called MaxDB, co-developed by SAP and MySQL when it was not yet a part of Sun Microsystems or Oracle, equipped with in-memory data storage and optimizations for row and column data stores for various kinds of OLTP and analytical processing. It is particularly tuned to make use of main and flash memory for data storage (instead of very slow but capacious disk drives) and to run on x86 server platforms. Being good Germans, HANA of course runs on SUSE Linux.
Sales of HANA, which started shipping last May, are running ahead of plan, and SAP wants to goose them further, while at the same time taking away some business from application and now database archrival Oracle at the same time. When reporting its financial results for the fourth quarter, SAP said it had hoped to book €100m in HANA sales in 2011 but that it had done €160m in business. SAP added that it had over €100m for mobile application products in 2011, slightly higher than expected and thanks in part to the Sybase acquisition.
SAP has announced that HANA is now ramped up as the database underneath its NetWeaver Business Warehouse data warehouse software, a process that started five months ago with early customer adopters. So now SAP can accelerate both data warehouses and analytics appliances based on commodity x86 servers from its HANA partners HP, Dell, IBM, Fujitsu, Cisco Systems, and Hitachi. Business Warehouse has over 16,000 customers.
In one test at the data centers at Volkswagen's Shanghai facility, a NetWeaver Business Warehouse query that took 20 minutes on a disk-based setup (using an unnamed database) took 45 seconds running atop HANA using memory and flash. The exact specs of the configurations were not divulged, and of course they matter a lot.
SAP wants to wedge HANA underneath all kinds of SAP application modules, and it also wants customers and partners to come up with new ways to use to help it sell HANA licenses. And so SAP has set up the HANA Real-Tune Fund and kicked in $155m to foster startups who want to develop real-time applications using the HANA DB database.
SAP also wants to displace "legacy databases" underneath its vast ERP suite, and has rejiggered its budgets and come up with another $337m dedicated to what it calls the SAP HANA Adoption Program. These funds will help cover the costs of SAP consulting services necessary to help customers make the jump from traditional relational databases (usually Oracle 11g and IBM DB2) supporting their SAP apps.
SAP also said that it has instituted an 18 month exchange program so if customers switch to HANA and are not satisfied (or adopt HANA the first time on new apps) they can switch to any other SAP product – that means Sybase, of course, which is a traditional relational database with some funky extensions and 30,000 customers worldwide.
Speaking of which, SAP also said that support for the Sybase ASE database underneath its Business Suite ERP stack will be generally available this month. SAP fully expects for customers to use a mix of HANA and Sybase databases for their ERP, with HANA doing the predictive analytics and some warehousing and Sybase ASE doing the grunt work where speed is not the issue and cost is.
And for those companies that need to manage very large data stores, SAP is reminding everyone that it also owns Sybase IQ, a columnar data store, which will share responsibilities with HANA. SAP said that is it working on connectors from HANA and Sybase IQ to Hadoop data munchers and will offer a "deeply integrated pre-processing infrastructure." No word on when this might be delivered.
Bootnote: Yup, MaxDB was created by MySQL and SAP, and nope, it was not an offshoot of MySQL but Adabas from Software AG. ®