Today I had the pleasure to attend a workshop at SAP in Walldorf in which I learned much about HANA, SAP’s high-performance analytics platform, followed by reading the excellent Ethan Jewett’s blog Thoughts and questions about the HANA announcementand browsing the SCN topic page on In-Memory Computing, which is also an excellent starting point if you don’t know what HANA is.
Great possibilities, much more to come
I’ve got to admit that even looking at the possibilities that are available with the current release, HANA 1.0, make me crave it badly:
- If you’re a BW customer, you can use HANA 1.0 similar to the BW Accelerator (BWA) to speed up reporting on your cubes drastically. (I’ve seen contradictory statements about this, so it could be that this usage will only be available with HANA 1.5 and BW 7.3, but I believe that HANA 1.0 can be used similarly to BWA and HANA 1.5 can replace BWA.)
- You can already replicate the data in your SAP Business Suite systems into HANA in real-time using the Sybase Replication Server or BOBJ Data Services.
- Add to that anything you store in non-SAP systems. Multiple ways to get the data from here to there exist, the most ubiquitous being flat file export and import.
- Data stored in a HANA database can be queried using BOBJ analytics tools, SQL, and MDX..
- Simple views can be built using a graphical editor, while more complex operations and transformations can be coded in a procedural language.
- HANA’s capacity for storing and handling data is spectacular, allowing any slicing, dicing, aggregating, and even calculations on top of trillions of records, or Terabytes of data, as a real-time experience. “Information at your fingertips,” as Microsoft promised us years ago, is becoming a reality in the Enterprise space.
And that is nothing compared to some of the things SAP might be doing with HANA further down the road (of which every trace was erased from my memory the very instant I learned that we were discussing them under a Non-Disclosure Agreement – but of course the roadmap concludes with HANA as the primary database of the Business Suite; also, check out the comments section of Ethan’s blog for more details).
Why haven’t I gotten my fingers on it yet?
So what stands in the way for me personally to start exploring the possibilities of HANA?
- First of all, no free trial download is available yet. (But if you have seen my SAP TechEd Live interview with Oliver Meinke and Mike Eacritt, you may remember that they were optimistic about an SCN trial download being available soon.)
- HANA requires massive hardware, the expensive professional kind that you need to install serious enterprise server software on; the kind so expensive that you need to get committees to decide to order them, possibly as a strategic investment. (As opposed to the kind of hardware daddy uses to explore NetWeaver ABAP 7.02 after work.)
- From what I understand, it has (currently) quite narrow hardware specifications, so while SAP works with multiple hardware vendors to avoid vendor lock-in, seems to support more than one meaning that even if your company has a big unused server standing around, chances are you won’t be able to run HANA on it.
So kindly put it into the cloud
I have never met a better use case for running software in the cloud. Evaluating HANA in an On-Demand offering similar to SAP’s BI On-Demand (e.g. http://bi.ondemand.com/explorer) brings out the key benefits of using software in the cloud:
- Pay as you go, because you only use resources as you go. No need to bind a lot of capital by buying massive server hardware whose resources remains unused whenever nobody’s slicing and dicing. In the cloud, unused resources are available to other customers, thereby reducing the cost for everyone.
- Get started immediately, because you don’t have to wait for a slow decision-making, purchasing and delivery process to complete. In the cloud, for example at Amazon Web Services (AWS) you can use virtual heavyweight servers on a per-hour basis for pocket money.
- Make yourself independent from hardware and their vendors by not having to deal with them at all and not making any investments into hardware. It would like calling a taxi cab instead of buying a luxury car.
Fig. 1: A good place for HANA
What might the HANA in the cloud service look like?
I imagine HANA in the cloud as follows:
- When you become a HANA in the cloud customer, SAP sets up a virtual AWS server with a pre-installed and configured HANA instance, including Sybase Replication server.
- Ideally, it is so user-friendly that you won’t have to use Remote Desktop, but can access it through a web page interface. (I don’t know if any of this is already reality but SAP can without a doubt build this if they want to.)
- If HANA needs to initiate a connection with your ERP system (e.g. to read database change logs for delta extraction), you can use the standard AWS secure IP tunnel solution.
- With a simple plug-in to your local database that enables a periodic push scenario, the tunnel becomes obsolete, and secure connections from your local server to HANA are achieved more easily with standard secure network protocols. (Again, I don’t know if any of this is planned or exists, but it looks like a good idea to me.)
- For the initial load, you can set up virtual storage at AWS, fill them with huge amounts of data, and just mount those disks from your virtual HANA server in order to import the data. Using the existing AWS’s virtual storage service allows you to use their standard processes for transferring huge amounts of data securely into the cloud by sending physical data storing media through mail or courier services (e.g. a trunkload of DVDs).
Cloud-based service and trial offerings can play an important role on the road to a wide adoption of the emerging next-generation in-memory computing platforms.
