Important Questions (with Answers) about SAP’s in-memory High-performance Analytic Appliance – HANA

I have just returned home after 2 conferences: SAP’s TechEd first in Las Vegas followed by WIS’s Reporting*Planning*Analysis in Orlando. It was a two-weeks speaker marathon for me: 5 lectures, 5 expert sessions, countless discussions over the meal/drink or simply at the corridor, but overall rewarding and enjoyable experience.

One of the hottest topics at the conferences was around SAP in-memory direction in general and about High-performance Analytic Appliance (HANA) in particular. As usually with every new technology, there is lots of noise coming from:

• Sales and marketing teams making bold statements without sometimes understanding the topic in details
• Analysts writing stories based on personal sentiment (positive or negative) for the company, product or trend
• Crowd (including me 😉 seeing the new path to get-rich-quickly and therefore trying to make the evidence of their presence before having something worth to say
• “Broken Telephone” effect

There is and there will be a lot of noise, so you need to filter what’s facade and what’s reality. The first wave of noise – after SAPPHIRE’10 – came and went quickly. It left behind new SCN forum “SAP HANA and In-Memory Computing Business Data ManagementSAP HANA and In-Memory Computing“, which dried up very quickly. Well, there was not much to discuss at that time. But now second wave is coming – SAP HANA is becoming a reality and is causing activation in the eco-system, including me.

During last two weeks I saw the full spectrum of feelings caused by the topic: from baseless hyperoptimism to arrogant ignorance to cynic skepticism. The same was with the spectrum of questions triggered: from relevance of persistency mechanisms for in-memory databases to the future of SAP NetWeaver BW. I think proper discussion should start from the basis. Below is my attempt to address the most important (IMHO) questions.

1/ What is SAP HANA and, more importantly, what is SAP ICE?

There is great thing SAP introduced with this latest technology: separation of the stack name – HANA – from the name of the engine – ICE – the core software component. Let us first look into the history though.

“BW Accelerator (BWA)” name was used for both – software as well as complete stack of software+hardware. This was a source of multiple misunderstandings and misinterpretations.

It was even more confusing with “SAP BusinessObjects Explorer, accelerated (BOEA)“, because now, although named after the front-end tool, the appliance still had BWA in the heart, even though the deployment scenario without SAP/BW became available. This caused pop-up’s of different names in different SAP materials, but I do not think there was any consistency in using them: “BO Accelerator”, “BI Accelerator” (yeah, welcome back, BIA :-), “Open Accelerator” etc.

Now SAP is introducing the new stack with the new software component in the heart, called “SAP In-memory Computing Engine (ICE)“. ICE is the in-memory platform consisting of database and calculation engines, and has been referenced in the past as “NewDB”, “HassoDB” or “BAE – Business Analytic Engine”. It is SQL ANSI 92 compliant in-memory database with full ACID transactions support, capable of storing tables in both column-oriented as well as traditional row-oriented form. Part of the SAP ICE is the calculation engine, which allows in-database calculations without moving data into the application layer.

In turn “High-performance ANalytic Appliance (HANA)” is the complete stack of SAP-certified hardware configurations with pre-installed ICE and some other software components. E.g. SAP HANA 1.0 will have Sybase Replication Server pre-installed.

2/ SAP HANA 1.0

The first release of the new in-memory analytics appliance from SAP is targeted for side-by-side deployments along SAP Business Suite applications. Please have a look at the top row on the picture in “The BW – HANA Relationship” post by Thomas Zurek. Sybase Replication Server – pre-installed in HANA 1.0 – will allow almost real-time replication of data from SAP ERP or CRM into ICE for further analysis. This real-time data from operational system can be mashed up in HANA with data from BW or some other 3rd party applications. For the latter data is not replicated, but loaded into ICE using SAP BusinessObjects DataServices.

Although I have not seen official dates for HANA 1.0 published, let me repeat what has been announced by SAP during multiple presentations. Ramp-up starts on November 30th of 2010 and go into General Availability when success KPIs for HANA 1.0 product are met (probably around late spring 2011).

3/ Business case for SAP HANA 1.0

SAP HANA 1.0 is intended for analytics solutions that require:

• real time data,
• high performance in-memory based processing,
• agile data modeling.

We need to understand that SAP HANA itself is like an empty database, and analytic solutions on top of it have to be:

• delivered by SAP or other providers in form of Vital View: SAP Announcement of Analytics Solutions,
• or, build in-house based on SAP business content or from scratch.

Obviously new technologies are opening new possibilities. Marge Breya, EVP and GM, Intelligence Platform and SAP NetWeaver Group, during her meeting with SAP Mentor Initiative at TechEd pointed out that HANA enables not only powerful analytics, but as well “algorithmic processing” like automated management of exceptions in streaming data.

4/ Beyond 1.0: HANA or … HOPA?

Although it is only version 1.0 of SAP HANA that is going to become available soon, most of the eyes are already looking into the future for the “on-top” deployment options – SAP’s in-memory as a replacement of traditional databases for ERP and BW – presented on the bottom row of the picture in Thomas’s blog. Although we saw a demo of ERP system running on in-memory during TechEd in Las Vegas, I will save my and your time i/o speculating on this topic right now. Hopefully we will return to it in future blogs.

I found that SAP does not always clearly communicate the difference in versions and their availability of new in-memory products, so some people expect they could run their BW or CRM on top of HANA “tomorrow”. Well, you need to be a bit more patient.

The only question I would have right now, is once we see the option of running ERP on top of in-memory database 2.0 coming, will it still be called HANA? I would expect rather something like “High-performance OPerations Appliance” or HOPA 🙂 Or may be simply ICE, because we would not need all the other surrounding software like Sybase Replication or SBO DataServices in the stack?

* * *

This blog is becoming too long, so let me save the question about how HANA/ICE are related to other SAP in-memory products for HANA 1.0 vs other SAP in-memory products.

-Vitaliy, aka @Sygyzmundovych

PS. Nov 6th 2010: We are living in the fast pace world indeed. I had to correct this blog the very next day after it has been published due to some important changes in HANA 1.0 architecture. And then few days later some other mention in the blog (not related to HANA directly) became the history as well. Viva La Evolucion! 🙂