Podcast: Debating the Value of SAP HANA
podcastlogo_jonerp.gif"Debating the value of HANA with four SAP Mentors on an international call...and discussing the burning questions for 2012". (ERP Lounge #22)"
Podcast Interview Date: December 3, 2011
Podcast: Listen Now!

Straight talk about HANA is hard to come by - but not with the group I pulled together for this international call. I have rarely looked forward to a call as much as this one, because I didn't know where the discussion would lead. I suspected we would have some significant points of debate and disagreement - I was not disappointed.

During this 60 minute podcast, you'll hear SAP Mentors and bloggers Vijay Vijayasankar, John Appleby, and Vitality Rudnytskiy talk HANA as only they can. It was a great discussion, and next time we'll try to get Harald Reiter in the mix as well (he was called away at the last minute).

The podcast covers seven main topics, in this order: 

- Sapphire Now HANA takeaways/Influencer Summit burning questions
- HANA, the cloud, and multi-tenancy
- Debate: is BW on HANA a "killer app?"
- HANA versus the Business Warehouse Accelerator (BWA)
- Our SAP HANA wish list for 2012
- SAP extending HANA to developers - success/challenges
- Epilogue: John's latest Bluefin HANA use case (gaming industry)

Editor's note- podcast links: Vijay Vijayasankar and Vitaily Rudnytskiy have both been blogging in detail on HANA. John Appleby has also blogged extensively on HANA on his Bluefin Solutions blog, his People, Process and Technology blog and also on his blog on SCN (check out John Appleby's HANA FAQ). Vitaliy also has a new blog called Vital BI. Other links: SAP Mentor Initiative. You can also check out SAP's In-Memory Home Page on SCN. Also: Experience HANA (with developer sandbox access)

Blog posts mentioned in this podcast: Vitaliy on HANA in-memory, John on "Project Orange" (BW on HANA), Vijay on real-time data and Red Bull on HANA.

Podcast timing: we taped this podcast the night before the SAP-SuccessFactors acquisition was announced. The cloud discussion at the beginning of the podcast should be heard in that light - none of us knew what was about to happen, but cloud was in the air. It is only a coincidence that all of us talked about the need for SAP to clarify its cloud message. :) Here's John's, Vijay's, and Jon Reed's take on the SAP-SuccessFactors acquisition.

If you want to subscribe to JonERP podcasts, get the The JonERP Master Blog and Podcast Feed. Or find Jon on his @jonerp Twitter feed. These podcasts are also included in the JonERP iTunes podcast feed

Podcast Highlights

1:18 The goal of the podcast - straight talk with informed HANA experts from the partner ecosystem, shedding light on Madrid and tough questions we'll have for SAP in Madrid in a week.

2:40 Vijay's reactions: my biggest news from Madrid was BW on HANA with Red Bull. Red Bull went in knowing what BW on HANA could and couldn't do. They took the stance that investing in HANA they would be able to influence SAP down the line - it was very smart of a customer to do that. Vitaliy: In the past what we've seen is that HANA is the major topic of TechEd. But once you entered the SAPPHIRE Now side in Madrid, most of the suits were spending time in the mobility pods. HANA was the number one topic for techies, but I didn't see as much interest in in-memory as I saw in mobility on the SAPPHIRE Now side. Looking to the Influencer Summit that is approaching, the focus won't be as much on HANA as on the cloud. So I want to ask SAP: is HANA still the central piece in this future technology vision, or is it getting mature and giving way in terms of attention to mobility and cloud?

5:02 John: Two major points to add: first point is a sense of maturity is happening, especially around mobilty- it's ready, and we need content. HANA is great as an analytics appliance, and it's getting there from a BW replacement perspective. There are 50 BW on HANA ramp-up customers, we've done our first migration and it feels like a great product. Second of all, looking forward to the Influencer Summit, SAP should look to make a big noise about cloud, put their sights to the cloud again. SAP didn't have much to say about the cloud in either Orlando or Madrid - they have something to say about cloud and are going to say it loud in Boston. Jon: I too am seeking clarity on SAP's urgency around Line of Business cloud solutions. I want to know about their resource commitment and what they're going to do.

7:06 Vitaliy - In the recent Gartner report on the future of HANA, they make an interesting statement that I didn't get any comments from SAP on, which was: when we are talking about HANA, forget about the whole appliance thing, and consider HANA architecture as a combination of cloud and in-memory. According to Gartner, this is where SAP is headed: is HANA evolving from the appliance model into the cloud deployment model and supporting SAP's on-demand vision? Vijay: Jon and I were in several meetings in Madrid where I just kept repeating the same question, which is: is HANA multi-tenant? We got varied answers. There were answers of "yes it is," And "no it isn't," and "no, but it will be." We talked to pretty senior people at SAP and didn't get the same answer. John: I'm going to defend SAP on that one a little bit: we asked it to the horse's mouth and he said very categorically, not now, but it will be. Vijay: but if you get six answers, that's an issue.

John: It's a difficult question; the concept of multi-tenancy and HANA is difficult. If you draw it back to Vitaliy's question about the cloud, SAP has to build out a hardware platform in the cloud, they have to create virtual public and private clouds for in-memory. You could argue that HANA is already multi-tenant if you want o see it that way, as you can run multiple databases on HANA in some scenarios. Jon: it's a tough question we were asking, but I can vouch with Vijay that these folks were not on the same page with their answers to HANA and multi-tenancy.

Jon: On the plus side, there was no philosophical defensiveness from SAP about the importance of multi-tenancy. I was really glad to hear that, that question used to be an open debate inside SAP. Oracle is now posturing on the same issue, saying multi-tenancy is not such a big deal. Hopefully by the spring we'll hear a consistent answer on HANA and multi-tenancy - I'm willing to give them another six months. Vitaliy: from a technical side, with HANA database code, if you look into it, there are many settings you can do per tenant, so there is some thought of multi-tenancy already in HANA, the question is whether HANA ready to support the multi-tenancy in the cloud.

13:10 BW on HANA - Jon to John: your  project orange blog post was maybe the most optimistic piece you'd ever written on HANA. Why do you think BW on HANA is a killer use case? John: let me frame this with the business problems that 14,000 BW customers face. They face challenges getting data into the environment in time for the business users that need it in the morning. They have challenges with agility, getting in the data quickly and making changes. The customers are also typically frustrated by the response time and performance of the BW system. For me, Project Orange resolves all those three things with one stroke. Load times are faster; you can be more agile by collapsing the layers, since you only need one layer and a semantic layer on top. And the in-memory leads to faster response times. So the BW on HANA questions is always a question of maturity. When SAP made available to me the BW on HANA software available in late October, what surprised me is that it worked the first time. It's stable, it's fast, and it was delivered to promise. That maturity made it interesting.

15:40 Vijay: I agree and I disagree. When I started moving from SP2 to SP3, even SLT stopped working. I actually think the maturity is quite poor. I also stopped on these internal tests for now, given vacation, and will start again in January. For me, it was the sheer amount of time it takes in a real customer situation to get BW on HANA working, it's not the "do it in a weekend" time I expected. Customers are on other OSes and databases, most are on BW 7.0 and they need to go to BW 7.3, and the OS DB needs the next version and so on. There's a lot of work before a good-sized BW system need to be done to get into BW on HANA. I’m not sure SAP has given a good enough explanation, it actually takes more time than we thought. John: I agree and disagree; on the one hand, SAP has positioned the Red Bull environment where it's a light house where it migrated to HANA in a weekend. While that might be true, we mustn’t underestimate the regression testing and the importance of the BW environments and versions. In our case, we had a much easier time. We had a project where we were on BW 7.3, that's the minimum required version for HANA, but it was on Microsoft, and it migrated in with very few problems.

18:46 Vitaliy: you are both right. There is some prep required, if you're on Oracle, you have to go to Oracle 11G for example. There is also some effort required on regression testing, HANA is introducing the new technology to the mix - without extensive regressions testing, it isn't possible. If I look into the OSS Notes for BW on HANA, one month from the project being announced, there are 90 notes to be implemented. This changes the ramp-up testing, so there will be bits and pieces of BW running on HANA, but besides all of that, I will agree as well that BW on HANA is going to be the killer app, it is going to be that simple. You almost ask "Why did it take so long for SAP to do something, especially with BW and performance issues? 

20:54 Jon to Vitaliy: what about BWA, didn't that address some of these issues? Vitaliy: BWA filled just a part of that gap. BWA is a good example of what we should expect going forward with HANA. BWA was used for a simple aggregation for the most part, even in the most recent versions. BWA introduced much needed discussions with what you can do in-memory and what's in the ABAP layer. Right now, what we see if you are putting BW on the HANA database, what is interesting how much optimization has been introduced into BW 7.35 to optimized on HANA. There is still too much calculation in the application layer. More of the BW calculation engine will be moved into HANA DB. With BWA, it was not all that easy, with HANA it will be much easier to go that way. I want to know from SAP developers: is it possible to do it non-disruptively, moving BW from one database to a HANA database. Normally you are just migrating BW objects, but SAP will be moving more functionality into the HANA database engine, so this will be an issue.

23:38 Jon to Vijay: Regarding your Red Bull blog post, do you have concerns that their system was only running in parallel? Vijay: Red Bull was very clear they invested in a long term view on HANA> They are running in parallel, they are looking ahead to Trade Promotion Management in ECC in a way that will benefit Red Bull, which I thought was very smart. About this "running parallel" thing: it's not a bad thing to do, many customers will do this with new systems. Let's give SAP some time on this, eventually they will get there. BW might be a killer app, I don't think so, but this won't take too long to figure out one way or the other. Even if only 5,000 BW customers move into HANA at $1,000,000 a piece, that's a lot of money. SAP's share prices should go through the roof, and if that happens, it will have a big impact. If 100 companies buy BW on HANA, maybe not so much. We'll find out in 2012.

26:38 Jon to the guys: The "game changing" rhetoric has been toned down in HANA marketing, but it seems to have given way to a lot of "big data" talk. Is HANA a big data solution? Typically we hear about 2 to 5 terabytes, which is not big data. When I think big data, I think Hadoop, and we actually heard from SAP on this topic in Madrid. John: I want to make a couple of points of clarity. First of all, I think there is no doubt amongst us that in-memory is liable to be an inflection point and to change the way that we relate to technology in the next 5-10 years. In terms of big data and HANA, the largest certified appliance is from Fujtsu and also IBM, which is a 16 node, 512 gigs server, That's at least 8 terabytes of in-memory, which is capable of running approximately a 40 terabyte Oracle or DB2 compressed database.

John: according to my definition of big data I think not, I'd say large it's large to medium data. If we're talking about Facebook, we're talking bout petabytes of data, and we're talking about transactional access versus aggregated access. What HANA is great at is aggregating large volumes of data. Vijay: it does worry me when HANA says it is big data when everyone takes "big data" to mean petabytes and above, but in SAP landscapes we are still in the terabyte world for the most part. Irrespective of the the rhetoric, HANA serves the purpose. With scale out solutions from multiple vendors, SAP is capable for handling the structured data in most SAP shops. I have philosophical differences on whether big data means an awful lot at the moment, from my point of view HANA is big enough data for most SAP data.

30:40 Jon to Vijay: in your blog post on real-time, you point out that the real issue is actionable information, not huge volumes you don't know what to do with. Vijay: and that problem is not solved by platforms like HANA. HANA doesn’t tell you what to do. The killer app part of HANA needs to be developed, meaning HANA runs on some value-added app on HANA tells you what to do with the data. If you expect users to write reports, HANA will be faster but won't have the real time value. Let's say you know your profitability is 3 percent instead of 4 percent, but what can you do about it now? If you need to wait a week to take action, then waiting a few seconds for the data to process doesn't do anything.

That's the right point, the whole idea for HANA as a solution came from requirements for real time business. I think it was quite bad that SAP marketing picked up this "big data" term that was existing on the market but means something completely different, compared to what HANA intends to achieve. HANA is not going to handle petabytes of data, but for example, the book I published on November 29 just became available in Google search today. So it took 4 days to be a part of Google's big data analysis. This is completely different than HANA, which is real-time data analysis. Big data today requires specific IT skills, whereas the purpose of HANA is to be agile BI and give the power back to the business user. If you look at the definition of big data and HANA, this move into one term causes confusion.

34:30 Vijay: In the HANA space, here's what I think should happen: two things are crucial. First, HANA security must be more sophisticated, and second, a metadata model built into the HANA layer. There is only so much SQL that can be handled at the database layer, so there is a lot of processing at the ABAP layer. A good example is: you go through a database of a billion records, but security only allows you 2 of them, so the benefit is lost. These things need to be pushed into the HANA level, the database and in-memory layer, and APIs must be exposed that developers can make use of without needing access to the underlying structures. That would help many more developers jump on HANA as a platform, because we can’t build business partners and vendors and so forth, but if SAP can do that with APIs, it will make things much more scaleable.

36:50 Vitaliy: My question is: what's next? "BW Powered by HANA" was a big expectation and now it's delivered, so what's next? We knew when this was going to be released, but we have no clue when to expect ERP on HANA. What I wish to see from SAP right now is: what is the roadmap from here. We do know some things about the applications - probably the most specific information is when certain RDS packages will be released that will use HANA. But then it's on to the platform level, for example we know there is no version management for debugging, so when is it going to be delivered. I still consider in-memory just a part of what HANA is built on. So those are the questions I have at the moment.

39:15 John: For me, this is about the right product at the right time, and I've had the sense that since SAPPHIRE Now, we've been very much headed in that direction, There's been an engagement with partners and we've had the ability to get that right. For me, if SAP really gets it HANA right in the next year, it would be how it enables its key and core partners to take the technology out in a way that gets it stable, ready, and fast. Which will make for happy and successful customers, and everyone wins. If we can make that happen on the HANA platform over the next year, that would be a resounding success.

40:20 Jon: We're still getting the occasional hyperbole-filled blog post around HANA from SAP, but we're seeing more customer stories and use cases now, which is good.
two things: COPA Acceleration - we missed out on a panel with customers, there were two live COPA customers in Madrid, and these are the kinds of Business Suite focused applications that can solve specific problems. Next year, we're going to see more applications that can address specific problems, but SAP has to figure out how to include partners and smaller development shops in this process, and that's been hard for SAP historically. However, they are making progress. SAP has figured out that it can't build it itself, so how are they going to include partners. Will they alienate them by building their own stack, or empower them by giving them sandbox access? SAP has been making real strides on this: 500 developers on are on Experience HANA sandboxes now, but it's just the beginning. Whether they can really make that available to developers to build apps on will be a big story for next year.

42:50 Vitaliy: It was a bold move on SAP's side to make this sandbox available for developers, I've had people next to me in the office who didn't get access to the internal HP HANA box but they got access to SAP's box and they are spending time pushing the system to its limit and testing its capacity. It is really helping these guys to get the real experience with the HANA box, and at the same time, this is the great feedback channel through SAP. If you look at in-memory data management forum, I cannot even follow it, you need HANA interpreters to follow the thread happening there. That would be a good place for unstructured HANA analysis based on forum activity.

Vijay: I do second that to a great extent. It was a bold move and it was the right move for Vishal and team to get it working. They did it quickly and efficiently - big kudos, they totally rocked it. Vitaliy: On the one hand, HANA is about extreme optimization which requires collapsing layers to eliminate any layers of abstraction, on the other hand, making HANA available to the ecosystem is about making open interfaces and letting other folks plug into your system and integrate the system with others. I think it will be a challenge in the future between driving optimization to the max and letting systems be open on the other.

47:10 John's HANA project: We just completed a project for a company into gambling. Suppose we had a smaller number of customers when they lose a lot of money or they make a lot of money in a small period of time. We put in a solution with SAP that involved capturing insights using the Event Insight platform, and we're making decisions based on information in HANA pushing real time offers. What are the offers we can make to those customers who are losing money a little at a time, supposed you can analyze it in real teal time, a know what's happening right here right then. There are agents with iPads using the Sybase Unwired Platform, and they can see the offers we should be offering to customers in real-time, so we put together a solution based on that which has gone live in the last week. As far as I can tell, it's the first in the world where HANA is used as a real transactional platform for this kind of thing.

It seems indicative of where the market is going. five or ten percent of their customers make 90-95 percent of their business, and treating them like individuals and dealing with them in a certain way will help them tremendously. Example: suppose you have somebody who is losing money slowly over a period of time and suppose they leave the table, suppose you can say, "Well I can see you're a great customer, please let us give you some complementary vouchers to keep playing, come back to table now.

52:35 Vitaliy: John provided an excellent example, and 2012 will be year of HANA stories, not just testimonials but deeper stories as more customers dive into use cases. I expect we'll be going deeper into the technology, and how the tech supports the solutions people are trying to implement. 2012 shoudl be the year of the HANA developer.