It’s over, no more sessions at the Annual Conference. No more swag at the vendor fair. No more questions all with the same answer. Back to the cubicle on the morrow. And what have we learned, class?
BI120 : Business Benefits Gained by TOMS Shoes by Migrating SAP BW to SAP HANA
Thu., 08:00 a.m. – 09:00 a.m.
As I reviewed my planned schedule on Wednesday night, I realized I had picked 3 choices for Thursday at 8:00. This happened as a result of the scheduling fairy’; when I went over the available topics I did not look first at the time slot, I looked at the content and at the speakers. It’s only after that first, second and third pass are done that I’d normally look for conflicts. Years ago I learned to leave more on the plate than I thought I could digest, as sometimes the speakers or the material is not what I expected. And I might even choose one session based on the room.
So I tweeted my conundrum. To get up early? Yes, of course.
Looking over my
The speaker from Tom’s Shoes first described the BW landscape. The details are in the slides, so I’ll just say their use sounds fairly typical, from the ETL layer to the SAP moving parts to the spectrum of data presentation techniques. The topic is HANA migration, so the end user views such as BOBJ 4.1 weren’t supposed to be functionally affected.
One major letdown to the story was their relatively new BW installation. In other words, not a lot of baggage that needed to be unpacked and repacked, thrown in the dumpster, or ripped out and rebuilt. Lucky them!
The company ran the migration in a series of sprints, with sprint 2 being the HANA migration. I’ll skip the technical details, as other than one bug found, the system move seemed unremarkable. The audience asked about the before and after database sizes, with the only clear answer being a 0.5 TB target system,
More interesting were the data load and reporting improvements. These are the advertised gains for HANA technology, and this is not a new subject in 2015, but it is surely helpful to hear from actual customers who have “been there done that.” I could not help but be impressed with the corporate responsibility posture Marcella shared. So what are the business benefits?
First, data loading is simplified. And faster. Good to know.
Second, data reloads are possible on an ad hoc basis, which typical nightly or weekly batches would not permit. I asked how these loads were automated, and how they were controlled (we use an external scheduler tool to coordinate these steps in our BW systems). Tom’s used the internal SAP scheduler only. When reloads are requested, it seemed to be more of a “trust basis” as they have relatively small teams that can communicate informally and quickly. With a much larger shop, I’d expect these loads to both be in a workload automation solution and to have a formal ticket request process (partly to satisfy auditor types, but also to communicate formally with approvals recorded.
Sales reporting is much faster. More data process in less time.
A sales report which prior to HANA included 2-3 months of data at a time and took 5 to 10 minutes to process, after HANA now can cover 2 years of data, and finish in 20 seconds (it says so on the slide below – you can’t make this stuff up. well, you can, but you shouldn’t).
Other happy news is the system seems very stable, with the above comment “fewer load processes means shorter load windows and opportunities for error.”
The survey at the end of the session, in the conference app worked for me.
Survey says… Um… No survey. But not for another attendee, who had no menu to record any feedback. Disappointing.
I heard in the hallways about other data faults (my gripe was the vendors not being actively shown in their booths – who remembers random numbers for more than a few seconds?). Not to mention I expect the survey return rate to be under 1 percent of the attendees from my experience and from the conversations with speakers later in the day who had no idea about the audience survey!
Walking through the show floor I spotted another demo of the latest supply chain planning software. And I’m posting this to make Somnath Manna jealous. 😆
Later, after visiting several partner/supplier booths, I stopped in the CA (Computer Associates) booth. That’s Dan Shannon on my right and Kelvin Levels on my right. Dan manages a development team for products including scheduler interfaces to SAP. We’d never met in person until today, though we’ve had numerous conversations about future code and current bugs. Nice chatting with him about things I might not bring up on a conference call. Of course they want me to present; I said sure, as long as it’s not in Las Vegas.
SAP Mentor sighting – Srini Tanikella – who was also an ASUG volunteer for years.
I stopped into this session during a lull in the action (hah) – it covered using CRM as a master data source, using synchronization techniques to get the data coherent. Here’s a shot of the agenda tool that tipped me off:
Interesting. Not something I can claim as a “win” back in the office, yet it’s always fascinating how different corporations leverage the same base software in creative ways. Oh, wait, this isn’t “simple”. Bzzzt.
Mentor meeting with Aiaz Kazi. Mainly about startups in the SAP world. Also about connecting those fledglings to big customers (like where I work). More on that if I am cleared.
Take my book. Please.
(apologies to Henny Youngman)
Get your tootsie fruitsie memory book here.
(apologies to Grouch and Chico Marx)
CP2053 : PepsiCo’s Manufacturing Planning Migration to Production Planning and Detailed Scheduling in SAP Advanced Planning and Optimization
Thu., 02:00 p.m. – 03:00 p.m.
Good session on production planning software and methods. Never once mentioned the H word, either.
BT1614 : Migrate to SAP HANA: Perform Database Migration Like an Expert with a Downtime-Optimized Approach
Thu., 03:30 p.m. – 04:30 p.m.
This one is more up my alley, as a DBA emeritus (one of my self proclaimed titles). System migration from one platform to another. With the target system having higher capabilities than the source system. Also using standard SAP tools – R3Load, SAPInst, etc. This reminded me a lot of the Unicode conversions we did moving platforms several years ago (see, e.g, Unicode – Episode 1.000: The Final Chapter (Updated 18-Feb-2008) ).
What is different here is the use of shadow tables to move large volumes of data while the original source production system continues to operate. This technique was pioneered years ago for SAP upgrades when customers could not tolerate the downtimes required when large objects were created or modified. The process uses change pointers so that any live updates can be tracked and committed to the targets system.
In fact, Unicode conversions can be included in this, meaning two separate upgrades (and their associated downtimes) can be rolled into one. There’s always the question of whether making two changes at once is too risky, so that is up to the customer.
There are caveats on the minimum source system version, but it sounded like this is less restrictive now than in the first versions. We found that to be the case when Unicode became a “requirement.” We decided to wait until others had tried the conversion and then demanded more efficient techniques from the vendors (both SAP and database). When we did Unicode, we used an Oracle tool (“O2O”). I’m not sure we could use that now (joking!).
An SCM system with Live Cache can’t be moved to HANA with this single pass DMO – Database Migration Option; it needs a two step process.
I asked how long the shadow table copies could take – hours, days, or weeks. Of course “it depends” on the data volumes though days is the most likely time frame. You’d presumably have some levers that could be pulled to slow down the copies if you decided your source system wasn’t beefy enough to handle the added load. Most of the time, companies would not want to add resources to a system destined for the scrap pile in short order, unless it was in a virtual container that had reuse capability.
The DMIS reader/writer processes shown above capture the deltas in the source system and apply them to the target system to keep the data synched. Lots of tools do this for different conversion needs, so I don’t consider this highly risky.
Good audience discussion about post-copy steps — this is identical to what happens in a production to quality copyback where the data are identical but configurations need to change to handle the new environment, like DNS/IP, system connections, etc.
Overall, a good presentation, and just technical enough for the audience. Except for that one guy that hit the wall, I mean.
OK, that’s a wrap. I got a Windows blue screen of death minutes before the end of the session. Wasn’t quick enough with the camera to capture it, though. A fitting end to a week of ups and downs. See you next year. (maybe)