It’s the last day of SAP’s enormous SAPPHIRE NOW 2013 conference here in Orlando, and the day opens with Hasso Plattner, one of the founders of SAP who still holds a role in defining technology strategy. As expected, he starts with HANA and cloud. He got a good laugh from the audience when saying that HANA is there to radically speed some of the very slow bits in SAP’s ERP software, such as overnight process, he stated apologetically “I had no idea that we had software that took longer than 24 hours to run. You should have sent me an email.” He also discussed cloud architectures, specifically multi-tenancy versus dedicated instances, and said that although many large businesses didn’t want to share instances with anyone else for privacy and competitive reasons, multi-tenancy becomes less important when everything is in memory. They have three different cloud architectures to deal with all scenarios: HANA One on Amazon AWS, which is fully public multi-tenant cloud currently used by about 600 companies; their own managed cloud using virtualization to provide a private instance for medium to large companies, and dedicated servers without virtualization in their managed cloud (really a hosted server configuration) for huge companies where the size warrants it.
Much of his keynote rebutting myths about HANA — obviously, SAP has been a bit stung by the press and competitors calling their baby ugly — including the compression factor between how much data is on disk versus in memory at any given time, the relative efficiency of HANA columnar storage over classic relational record storage, support on non-proprietary hardware, continued support of other database platforms for their Business Suite, HANA stability and use of HANA for non-SAP applications. I’m not sure that was the right message: it seemed very defensive rather than talking about the future of SAP technology, although maybe the standard SAP user sitting the audience needed to hear this directly from Plattner. He did end up with some words on how customers can move forward: even if they don’t want to change database or platform, moving to the current version of the suite will provide some performance and functionality improvements, while putting them in the position to move to Business Suite on HANA (either on-premise or on the Enterprise Cloud) in the future for a much bigger performance boost.
HANA is more than just database: it’s database, application server, analytics and portals bundled together for greater performance. It’s like the new mainframe, except running on industry-standard x86-based hardware, and in-memory so lacking the lengthy batch operations that we associate with old-school mainframe applications. It’s OLTP and OLAP all in one, so there’s no separation between operational data stores and data warehouses. As long as all of the platform components are (relatively) innovative, this is great, for the same reason that mainframes were great in their day. HANA provides a great degree of openness, allowing for code written in Java and a number of other common languages to be deployed in a JVM environment and use HANA as just a database and application server, but the real differentiating benefits will come with using the HANA-specific analytics and other functionality. Therein lies the risk: if SAP can keep HANA innovative, then it will be a great platform for application development; if they harken to their somewhat conservative roots and the innovations are slow to roll out, HANA developers will become frustrated, and less likely to create applications that fully exploit (and therefore depend upon) the HANA platform.