GRC
HR
SCM
CRM
BI


Podcast

 

Dr. Pepper Refreshes Its BW System with SAP HANA

Part of The HANA Effect podcast series

December 10, 2015

Roopa Amin, BI Director at Dr. Pepper Snapple Group, joins SAP’s Jeff Word to discuss Dr. Pepper’s migration to SAP Business Warehouse (SAP BW) on SAP HANA. This podcast steps through Dr. Pepper’s complete project, including:

  • How Dr. Pepper Snapple Group decided to switch its SAP BW system over to BW on HANA
  • The business and technical benefits of the migration from a slow 20TB DB to a screaming 5TB HANA DB
  • What cutover weekend looked like, and business users’ reaction to the move

Listen to the podcast, and read a transcript of the conversation here:

Jeff Word, SAP: Welcome to another episode of The HANA Effect. I’m your host Jeff Word from SAP, and I’m joined today with one of my favorite customers, Dr. Pepper Snapple Group. We’re here today with Roopa Amin. She’s the director of BI. Welcome!

Roopa Amin, Dr. Pepper Snapple Group: Thank you, Jeff. Thank you for coming here.

Jeff: We want to talk a little bit today about the BW on HANA that you guys just went live with a couple of months ago. So, take us back in history a little bit and talk to us about what led up to that decision to switch your regular BW system over to BW on HANA.

Roopa: We were actually implementing a project called Capstone. And this project was to add some case sales data into our existing BW system. Case sales, at a very basic level, is the sale of our product to a retail store at a SKU and day level. The sales at retail stores are of course done by Dr. Pepper. We sell our product, but so do other bottlers. And we take our SAP data from our ECC 6 system and combine it with this external data, and do a lot of reporting. And the case sales data gives us some insight into who are our most effective bottlers, and what funding models should we use, what sort of displays should we fund, and what type of promotions. As we were implementing this project, we found that our data volumes were much higher than we’d initially projected, and therefore the data processing requirements were much higher, as well. So we started looking at HANA to see if we could optimize our data load times. We did two proofs of concepts, one with another vendor and one with SAP, to really understand which product would be better for us in the long run to improve our data processing times.

Jeff: What size data volumes are we talking about?

Roopa: When we were cutting over to production at that point in time, our database was about 20 terabytes. If you exclude all the unused space and indexes, etc., it was probably around 13 terabytes that we actually migrated.

Jeff: And what did you end up now in HANA with the compression?

Roopa: It’s about 4.5 terabytes.

Jeff: So about 5x compression. That’s actually pretty standardized. The data you talked about, this case level data, is very, very granular, right?

Roopa: Yes. So we used to hold the data in our legacy systems at a chain and a month level. And now we’re bringing it down to a store and day level. And this is not just our internal data, but it’s data from external bottlers that we pull in.

Jeff: So you can really see all the different products flowing into the retail stores.

Roopa: Exactly, right.

Jeff:That’s amazing. And what was the result when you did that proof of concept?

Roopa: We saw a significant difference between the other product and HANA. HANA gave us the performance that we were looking for. There were definitely some places that were 10-12x faster than the other product. And then, there were areas like DSO activation that you have in BW that we couldn’t do in our current environment. And we certainly couldn’t do it in the other POC. But on HANA, that flew. It was amazing.

Jeff: What were some of the key decisions that you guys had to make as you were getting ready to embark on this HANA project for BW? Roopa: So some of the key decisions that we had to make were first of all, which hardware vendor do you go with? We have a very good relationship with HP. We have a hosting agreement already with them. So we decided to go with HP hardware. And then we didn’t have to make a decision on, did we want it on-prem or did we want to host it in Toronto with the rest of our hardware. Some of the other things that we had to think about was, do we want to patch our BW and if so, which version? Jeff: When you guys actually went live with this thing, what was the technical aspect of your cutover to the new HANA system?

Roopa: There are a lot of different migration options. One of the migration options is to have a zero downtime. We could not pick that as an option. We’ve got a lot of non-SAP interfaces, and the work involved in actually setting those up for two environments was going to be difficult. And so we decided to go with the downtime option. So one of the challenges that we had was to really make sure that we could reduce that to as low as possible. And so we used SAP Services to help us come up with techniques on lowering that downtime. It’s a bit of an art, but we were really successful at it. So that’s one of things that we looked at. We also worked with the Ambassador program to help us identify which version of the software we should go to. We made decisions on what patch level in BW. But in terms of the actual cutover, it was very uneventful. We had one small issue. Other than that, the actual cutover went really smooth. We were done actually sooner than we had anticipated and we were able to start our data loads earlier.

Jeff: It was ahead of schedule.

Roopa: Yes, and I think one of the reasons was we were going from a production environment to a production environment, so that made it a lot faster.

Jeff: So what was your life like over that weekend when the cutover happened?

Roopa: You know, you’re always worried about how will it go. But it was so smooth it was actually a little bit boring.

Jeff: So, let’s talk now, you guys had BWA for a few of your cubes. But now on Monday, when the whole company comes in, all of their stuff is now turbo-charged with HANA. What was the reaction of the business users before and then after? What were they thinking or expecting about HANA and then what did they really feel when they got it?

Roopa: And so, the users that actually didn’t have their data on BWA, they were just ecstatic. We had one person who wrote to me that said, I used to spend about 2 hours running a report, and now those same reports run in about 10 minutes, and I use the balance of my time to do the analysis. And then in terms of the rest of our users, what was interesting was, first, everyone was so happy that it had gone in so smoothly. And then in terms of the month-end close for our business users, I debated on whether I should email them and tell them, hey, this is our first month-end close on HANA. But I decided instead just to see their reaction after the close happens. So typically, what we do is, the month-end starts around in the afternoon and the jobs run, and late into the night, our business users are asked to log in and validate the data so that we have it available for our sales guys in the next morning. So instead, we started our jobs as normal in the afternoon, and we were done in the evening. And I got an email from one of our business users that said, this is amazing, you know, HANA really works. And she went and told our senior management — in fact, the president of our beverage concentrate business — how well the implementation had gone and that we were able to deliver our reports sooner.

Jeff: That is a pretty big applause and ‘atta boy.

Roopa: Yes, and we’re still working to improve it some more. We’re now starting to get rid of the cubes and move our reports to the DSO. So every month, we see a little improvement in the process. We’re continuing to condense our cycles down. Until now, we couldn’t even run both close cycles at the same time. So we have our ECC 6 data and this bottler case sales data. And we had to run one in the day and one at night. Now we’re shrinking down the times so we can run both at night and optimize our offshore resources to actually monitor those data loads at night and use our on-shore, more expensive resources to work with the business.

Jeff: We’ve heard about a lot of other success stories from a lot of other SAP customers. And the confidence that the business gets in that first project, of yes OK, HANA’s different, but it really delivers. This stuff really is as good as everybody hoped, probably sometimes even better than they expected. Then they start building on that and looking and saying hey, what can we do next? What can we do next? Where can we apply this new technology to improve other aspects of it? That’s what we call the HANA effect, right? So explain to me a little bit about what Dr. Pepper is going to be doing next in your HANA roadmap? Where do you think that this type of technology is going to have an impact going forward for you guys?

Roopa: We started our HANA roadmap discussions now. And so, our senior IT management is very interested in pursuing putting ECC 6 on HANA. And actually, the recent announcements about S/4 are just amazing. Everyone is very impressed by the whole scenario of reducing the data footprint by consolidating the tables. It makes so much business sense, it really does. And eliminating all of the checks and balances people have to do across all those different transactions. We spend so much time closing our financials, etc. So we’re really looking now into how we’re going to optimize HANA further. And so our next step is to really understand how we migrate ECC 6 to HANA and then really, subsequently, how do we implement Simple Finance, etc. So what we want to do is really look for good use cases to showcase how we would add business value as we implement this further.

Jeff: Now that you guys have had this massively successful BW project, you’ve learned a whole lot, I would imagine through that process. What kind of advice would you give yourself now that you’ve been through it, now that you know?

Roopa: Well, one of things that I wasn’t sure about was, what sort of skill sets would I need. It is more Basis intensive, so you want to make sure that you have some strong Basis staff on your project. Another thing that you want to make sure you account for is all the lead time and installing and setting up the hardware. You also want to think about all your things like, how do you want to handle failover on a node? Do you want disaster recovery? Do you want to mirror it in another data center? So all of those thing which are available with the HANA product, are choices that you need to go through and make decisions on.

Jeff: All of those are incredibly important things to think about, and they all depend, based on your situation.

Roopa: And you also have some non-business value that you gain. So, for example, we talked about the data processing times are reduced. We talked about how query performance is improved. But there’s also some other things. And some of those things are, well obviously, from a Basis perspective, the amount of time building indexes, doing restores. Restoring from production down to QA is so much faster. We did our first restore just recently, and instead of taking days, it took hours. It was our first time, so we were learning. But we really think we could do a refresh from production down to QA in about an hour. The other things that are really great are backup times. Our old database, it used to take us 14 hours to back up that database. So imagine three days a week, there’s a job running for 14 hours to back up your database. And it’s using resources that your business users are also competing with to run their reports. And if you run too many things in parallel, the system can tank. Now those backups run in an hour, even less actually. So backup restores, refreshes, DBAs not having to build indexes for you. And then just housekeeping things. Like, if you want to delete data. You know, you’ve got PSA data, you’ve got change logs, etc. Deleting those used to be difficult. And now we can delete data so much faster. It makes housekeeping tasks so much easier.

Jeff: What were some of the things that your technical people had to adjust to in order to take advantage of all the new stuff in the turbo-charged HANA world?

Roopa: Definitely from a developer’s standpoint, they needed to think about simplifying what they did. So before, we had all these layers that we used to build, and now we don’t. We also can allow for more granular data reporting. In other words, instead of having to summarize everything and stick it in a BWA cube, we can now bring it in at the lowest level and start reporting. From a Basis perspective, there’s a lot less work. There’s not as much work to build indexes, or, there’s no work to build indexes, etc. So the focus has now shifted to more of the housekeeping, keeping the system lean, doing housekeeping that one should have always been doing. And not spending time on trying to actually tune it. So there’s not as much time spending on tuning. There’s more just basic housekeeping and providing reporting faster.

Jeff: Basically squeezing more value out of that data, because you don’t have to do all of the stupid stuff anymore.

Roopa: Exactly. All the non-value-add stuff is gone.

Jeff: Well listen, Roopa, this has been absolutely phenomenal. Like I said, you guys are my favorite caffeine dealer. Love, love, love the product. And love the company. You guys have done a phenomenal job here. Can’t say enough about the success of this BW project. I’m just thrilled to see the momentum and the confidence that it’s built, so you guys can continue marching forward on that road to simple S/4 HANA, Simple Finance, and simplified IT operations.

Roopa: Yes, we’re excited about more implementations using HANA.

Jeff: I want to say thanks a lot for sharing your story with everybody on the HANA podcast.

Roopa: Thank you.

An email has been sent to:






COMMENTS

Please log in to post a comment.

No comments have been submitted on this article. Be the first to comment!


SAPinsider
FAQ