My day started with informal discussions on future of BI products. Mobile BI is of course a no brainer next big thing. Question is how is SAP going to tackle this? Webi is by far the most commonly touted mobile BI client. Xcelsius has the flash limitation on iOS, and even on Android – not all controls of Xcelsius work that well. Some time next year, when an Analysis client for mobile goes into ramp – there is an option to generate HTML5 layouts. If SAP does a good job on this, I would bet that this will start Xcelsius on its death march. Xcelsius is a great tool – but popular opinion is that it is not scalable for thousands of users. I have seen few hundreds use it, but not thousands. So I personally don’t know, and am willing to stand corrected. I will ask around more while I am here.
Next up is a question of how to best make dashboards etc out of ECC. The big limiting factor in my opinion is that universes do not work on top of ABAP data dictionary. And if you go directly against the tables at SQL table – you will lose out on security, semantics and most probably will violate licensing agreements with SAP and the DB vendor. The smart thing for SAP to do will be to allow universes to be built on ABAP data dictionary. This could single-handedly give an explosion to licenses sales for BOBJ. Any one listening at SAP from sales side? To check out – I spoke with Adam Binnie, the GVP for BI. Many thanks to Andrea Kaufmann for finding some time for me with Adam.
There is no solution now or on roadmap to solve the flash compatibility issue. Adam said SAP team is checking on this in a research mode. It is not an easy problem to solve – there is a huge userbase for whom a disruptive solution might be painful to the extreme. Plus SAP has a limited option on a problem between Adobe and Apple. Nevertheless, SAP owes the world a solution for this somehow.
Although I feel SAP has not given Xcelsius the love it needs in near past, Adam thought otherwise. I think the disconnect essentially is that whatever is driving the product roadmap, probably is not very well integrated with the ecosystem. Adam said he will look into it, and that by end of the year he plans to put some concrete actions in place.
On the universe on top of ABAP data dictionary issue, it did not look like a priority for SAP. Adam thought this might be because multiple semantic layers (as in ABAP DDIC, and Common Semantic Layer from BOBJ both) might be against SAP’s design principles. Again, he agreed to check with his team and give me an update.
Of course no day passes in SAP land for me sans a thought on HANA. if we fast forward to future – the big big thing is ECC running on HANA. Is this the final frontier? I am not too sure . Over the last decade or two, SAP has made a lot of DML optimization already. Standard Business Suite behavior is to select once from database, hold it in memory – and then do all kinds of business logic in ABAP. So you could select a million records, filter it to hundred thousand with logic – then take it through Authorization checks to find what will pass, and end up with 10 records. If this logic is not rewritten completely – how much of benefit will SAP get by simply putting the tables in memory? And think of the re-write – can you replace the ABAP logic with a bunch of database stored procedures ? Can you find a clever way of passing authorizations to hana at database level and not let ABAP do it? Will this mean HANA gets an ABAP interpreter? Too many complex issues to be solved to make it work efficiently. So I raised this with SAP’s Deputy CTO Sethu Meenakshisundaram and a member of his team, Frank Samuel. Here is what they had to say.
SAP has a vision on how this would work going forward. The key to that is building a good foundational meta model in HANA that can represent the business objects in ABAP. Once DDIC, Security etc are modelled, then a standard API set can be used by ABAP and non-ABAP systems to do complex processing with HANA. But given the sheer amount of work, this will take some time to complete. A consistent set of user friendly APIs is a tall ask for SAP in my opinion, based on what I have seen in past. Maybe they will do better this time.
Jon Reed brought up the question on multi-tenancy in the meeting. And I had posted this question on my blog earlier. I was quite pleased that Sethu had read my blog before hand. So according to Sethu, HANA is built as a multitenant model, except it is not used as such. Apparently, this is a simple thing for SAP to flip the switch. This is very different from what I have heard before, and I mentioned it to Sethu. He agrees that communication on HANA needs to improve from SAP’s side to minimize confusion. In any case, SAP not only needs to flip that switch, it should also build apps on HANA that are multi-tenant. And Vishal should mention it in his keynote to make sure the world is not confused, unlike what happened with ORACLE. BI OnDemand is a good start in this direction, and I expect more such things to come up in next few years
SAP has a grand vision on how the in-memory platform will help business. And i truly like what I heard. Sethu thinks Supply Chain is an area where SAP can do wonders with this, and I agree. S&OP for example is an area where I have heard many customers request more of SAP. However, vision and execution needs to match. SAP’s execution on HANA is not exactly stellar so far in my opinion, but I guess they will get there soon enough.
Next up was a meeting with Bjoern Goerke who is the Corporate Officer in charge of Technology and Innovation Platform at SAP. As always – a terrific meeting. I asked the same question on HANA multi-tenancy to him. His answer was more tentative than the one I got from Sethu – “not now, but we are working on it, and will get there soon”. While he agreed on the need for a good meta model on HANA and a framework for all the aspects like security, user provisioning etc – he could not give out any time frame for it. “One step at a time” was his answer. I respect that, although I think these are all things SAP should get out to the ecosystem as soon as they can. Technology should not be a huge limiting factor here – ABAP DDIC itself is stored in tables, and parts of this concept is already in use by other parts of SAP like BW replication etc. SAP should do all they can to improve speed to get these things out of labs to the ecosystem. I asked Bjoern if BOBJ universes will be built on top of ABAP DDIC, and did not get a clear answer.
I did provide some feedback that HANA studio is not exactly a fine artwork, and got some assurance that it will improve constantly. SAP is moving all the developer studios to eclipse, which is the right thing to do. ABAP on eclipse is already available to see in the booth here. Apparently 7.3 Netweaver is doing great, and has 300+ live customers. A lot more info on neo and ByD etc were discussed, and between Jon Reed and Dick Hirsch – I bet they will post their views after the meeting.
I would be terribly remiss if I did not mention this – Stacey Fish and Andrea Kaufmann from Mike Prosceno’s team deserve big time kudos. They are not only super helpful, they also try their best to give us as much information as possible so that we can form a more educated opinion before we blog. If they worked in my team, I would be writing performance bonus recommendations for them now. I cannot thank them enough