Thoughts and ideas on HANA...definitely not at HANA speed.

[GoLaSalle]

(This was originally posted on my personal blog as a bit of 'public therapy'. But I thought that maybe the community might share the same questions, so I've cross-posted it here on SCN to see if anyone wants to discuss.)

I do not profess to be a HANA or 'big data' expert- I'm primarily an Analytics architect in a relational world. But I'm a geek and HANA intrigues me. That curiosity has definitely been piqued in over the last month, starting with the launch of HANA Enterprise Cloud right before Sapphire/ASUG Annual Conference. And of course, the subsequent references during and after the conference.

The overall picture has begun to become a bit clearer through the murky waters. I've been using these blogs as my primers:

Dick Hirsch: Understanding The HANA Enterprise Cloud- An Initial Whiteboard

Aiaz Kazi: Evolution of the SAP HANA Cloud Platform

Matthias Steiner: What the HEC(K) is Going On?

Vijay Vijaysankar: A Faster Horse Just Won't Cut It Anymore

Since I do not profess myself a HANA guru, I have NO intention of writing anything to the level of the above. But from the outside looking in, what does all of this movement seem to say? SAP is pushing forward with it's 'all in' HANA strategy, correct? Given the technical effort, marketing push and recent news- that's pretty obvious.

On that note, I've heard some interesting discussions regarding the latest HANA Enterprise Cloud and HANA Cloud Platform announcements. Some SI's/partners/consultants think that SAP might be moving into some of their key spaces. Others think SAP is taking a logical step by offering these services and challenging partners to step up their game, while reaping the benefit of solidifying the HANA market with a level of expertise that currently SAP (and a few select partners) can currently offer. Personally, I think that only time and actual customer purchasing will really tell. Once HANA becomes a commodity, does SAP bow out of the cloud hosting model? These are probably questions that don't have clear answers now, only in the future. But it doesn't make them any less fun to discuss.

Turning to my home turf, what about Analytics? There has been a lot of discussion back and forth about the future of the semantic layer. Today, each semantic layer has a necessary use case/requirement. But what about in the future? Does HANA survive 'the Game' and take 'the Prize'? There can be only one. </end Highlander references>

Can (or even should) SAP really position HANA as the one 'semantic layer to rule them all'? Companies who can't/won't buy HANA as a database may be open to purchasing HANA as a 'BI Accelerator'. At this point- is HANA any different than installing the 'accelerator' server for other vendor's Analytics tools? Can I can model my data views in HANA and since all the SAP BI tools connect to HANA natively, HANA BI Accelerator Edition is the semantic layer? In the perfect world of my brain- this works. Of course, the reality is starkly different. Licensing, fees, server installs and configurations, designer tools...the list could go on and on.

Think about the possibilities. One semantic layer: a common business layer that brings a single vision of your data. (Yes, yes- I am sure that some may call this data federation. But in some enterprises, that is a 'dirty term' and not easily implemented.) Perhaps that's too Utopian of a vision. Disparate systems, databases, security requirements- that is the reality in most enterprises. What if I could model my HANA layer to connect to my big relational warehouse and a smaller departmental database; bring the data back into HANA and make it available for my Analytics tools. Once my query is done, the dataset is blown away- no persistence. This makes HANA an Analytics 'Appliance' or 'Accelerator' and not an 'in-memory' database. Are there use cases for this kind of implementation?

Perhaps another option is to break out the database from the application engine and use the engine only as your new semantic layer? Instead of connecting to a UNV/UNX/etc, you connect to a 'model' that lives in HANA. No data is brought back, HANA acts as the semantic layer to your data. SAP or agnostic datasources? Doesn't matter. BI developers now have one tool to connect and model. Use cases for this particular implementation may be easier to come by, but is it actually doable?

Did you notice a lot of questions above? Me too- that's why I wrote this blog. I think that there are still untapped markets and possibilities regarding HANA. Does SAP continue to drive the market to a place it thinks and hope that when it gets there, the price and time to production are commiserate with the public perception (and reality)? Or can SAP let the marketplace drive where to go next?