How Accurate Was John Appleby on HANA Replacing BW?
Executive Summary
- John Appleby made bold predictions on HANA.
- How accurate was his article on whether HANA replaced BW?
Introduction
John Appleby’s article on the SAP HANA blog was titled Does SAP HANA Replace BW? (Hint: Still, No.) and was published on April 17, 2014.
Our References for This Article
If you want to see our references for this article and other related Brightwork articles, see this link.
Notice of Lack of Financial Bias: We have no financial ties to SAP or any other entity mentioned in this article.
The Quotations
Simplification of the SAP Landscape?
Nearly two years ago, I collaborated with my good friend Steve Lucas on Vishal Sikka talk about the simplification of the SAP landscape. For instance, SAP CRM contains at least 40% duplication of data from SAP ERP, so if you were to combine both systems into one platform, you would see a significant reduction in overall storage and an association simplification in IT landscape. In addition, SAP BW is obviously a duplication of data, when it comes to SAP data stored in another system.
It seems likely that in future revisions of SAP BW, SAP will allow a linking of information from SAP BW into the Business Suite, so real-time data won’t need to be duplicated into BW. This will also provide more real-time access to data warehouse information.
So I wouldn’t say so much that BW makes things more complex, but rather that it allows you to encapsulate the complexity of your business in one place – including SAP data from ERP, CRM, SCM and other systems, plus whatever other kinds of data you need from other places in your business.
Curiously, Appleby lists Steve Lucas and Vishal Sikka as good friends. The curiosity is that all three of these people are our lowest ranking sources on SAP. We covered Steve Lucas’ terrible information on HANA in this article, Analysis of Steve Lucas’ Article on What Oracle Won’t Tell You About HANA. Something else that is an issue is that Appleby states that he is not compensated by SAP, but he is excellent friends with these top people at SAP, and along with John’s position at the time, all three had a strong incentive to push HANA. Let us say that Appleby found something unflattering about HANA, would he tell his readers?
As for the duplication of data, this is commonly the case. But there has been a long term reason for this. And even in 2019, there is still duplication of data. It is also inefficient to try to make everything real time, and the needs of reporting are rarely actually real time. That is, day or multi-hour delays work perfectly well for SAP and non-SAP companies.
Implementing a New Data Warehouse?
If I’m implementing a new data warehouse, should I implement it in HANA Enterprise or BW on HANA?
In my HANA business, there has been a resurgence of data warehouse implementations and this is a common question. What I normally reply to this is that SAP HANA in itself is an excellent database, data platform and application platform. However SAP HANA on its own does not provide a number of the functions that are required in a complex Enterprise Data Warehouse.
For those functions, SAP provides the BW EDW platform – BW takes care of SAP security and authorizations, complex hierarchies, non-cumulative data for stock and many other capabilities. In addition, the BW team has worked hard to update SAP BW 7.4 for the HANA generation, with a set of technologies it calls LSA++, which simplify and accelerate the build and operations of an EDW.
In short – if you want to build an app, data-mart, data platform or database, then use HANA on its own. If you want to build an Enterprise Data Warehouse, then use BW on HANA.
Companies that want to use data warehousing and SAP use BW on HANA, not HANA Enterprise.
Still, Investing in BW?
So SAP are still investing in BW?
Yes – more than ever, in my opinion, with the latest BW 7.4 version. If you’re interested in the detail, there is a good Roadmap available, but the short version is that the BW team have invested to take advantage of the HANA DB. This means faster data loading, fewer layers of information, more agility. It allows you to mash-up BW and non-BW data (if you have a mixed license that allows for this) and real-time with batch. Plus, a lot of the tooling has been reworked. The BW guys have been hard at work and BW 7.4 is a real step change.
In addition, because BW uses a very low level of HANA calculation (lower than you can model in HANA Enterprise), for complex reporting requirements and query structures, BW on HANA can be much more flexible and performant than HANA Enterprise. If you’re interested in the detail, there is agood explanation by BW lead Thomas Zurek. Yes, in some cases, BW on HANA is more capable than HANA Enterprise.
In addition, SAP changed the license late last year, and whilst you should check license detail with your sales rep, you should find that your license allows you to load data into HANA via BW, but then use all the HANA modeling tools to build data-marts, if you so wish. This is very useful for the build of custom data-marts in BW on HANA, though remember, you lose some of the BW functionality by doing this so it must be done with care.
SAP was and is still investing in HANA. This is a bit odd if one thinks about it because a lot of what BW is, is star schemas (0r modified star schemas as those with appropriate technical knowledge state that SAP implements the star schema in BW incorrectly.
So the fact that SAP is still having customers use HANA and BW is a design issue and overall product vision issue because BW is designed to create structures that speed performance from a row-oriented database. But the fact is companies have invested hugely in BW, and they are not going to change their approach to data warehousing very much, even though HANA changes how BW should be configured.
Cold Data with IQ
What about cold data. Does BW on HANA support this?
Yes, and it supports it very well with the NLS connector for Sybase IQ. We find that for data warehouses larger than 5TB in Oracle, that a good portion of that 5TB is cold data that is rarely accessed. In this case, it doesn’t make sense to buy expensive in-memory hardware to store the cold data, and it makes sense to store it in a separate store. SAP allow this in BW via a connector for Sybase IQ for Near-Line Storage.
So your older, non-changing data becomes stored in IQ and is visible the same way in BW and this improves the TCO equation for HANA significantly. If you have a smaller BW system then this won’t be necessary but it is a great addition for large-scale systems.
HANA does not support cold data. SAP stated that all data should be moved into memory. The story changes that cold data is not kept in HANA (of course) but is placed in IQ. But IQ is not HANA. The question is whether HANA supports cold data. And another reason HANA doesn’t is that it is too expensive to store cold data in HANA due to HANA’s pricing.
Secondly, IQ has not been a popular choice to connect to HANA. IQ was from the Sybase acquisition but has seen its popularity steeply decline since the purchase, and it was never designed to be an archival system for HANA. This new role was created so that IQ would not be positioned as competition to HANA. IQ is a primarily similarly designed database to HANA that is far better designed and is far more stable — but not developed by SAP. Therefore, to protect Hasso Plattner’s ego and the overall false narrative of HANA being innovative, IQ was pushed to a secondary role. Finally, SAP has no demonstrated expertise in building archiving solutions. SAP customers should choose their archival solution independent of their HANA selection (if they were unfortunate enough to have purchased HANA).
Ahmed Azmi explains IQ.
“I don’t know what they did but with ASE you get a great RDBMS for OLTP and with IQ you get Columnar storage and compression (for archiving data). which gives you OLAP as well.”
The Sybase acquisition was in 2010. That was plenty of time to include parts of it in HANA. SAP has a top OLAP DB — but decides to position HANA as the primary combined OLAP + OLTP (adds rows in SP08). Then positions a column-oriented DB into an archiving DB for HANA. SAP stopped IQ from being positioned as an analytics DB and was positioned as HANA’s archival. SAP stopped mentioning Sybase because it goes against the fake story that HANA was invented by SAP and not some Frankenstein mix of other DBs as we covered in Did Hasso Plattner and His Ph.D. Students Invent HANA?
BW On-Premises or In the Cloud?
Should I implement BW on-premise, or in the cloud?
You should do whatever suits your business. I’ll be honest with you, most of the BW customers I deal with are big: 5-10TB Oracle or IBM databases, which require large HANA clusters of 10-20 HANA servers (including Dev/Test/QA/Prod/HA/DR plus growth), and right now, most of those customers are still buying on-premise hardware. For large-scale systems, it seems that on-premise may still have a lower TCO, and they also require a lot of network capability, which can be a concern in the cloud.
But I’m not sure this is indicative of the future and we are starting to see a shift with some customers considering large BW systems in the cloud – usually the SAP Cloud for large systems. In addition, we see a lot of use of the SAP BW on HANA Trial Edition which runs in the Amazon Cloud, and the Amazon Cloud is also now supported for smaller BW systems, which is also an option. In addition, there will be partner clouds by companies like EMC and T-Systems in 2014, and I’m expecting this to be a serious consideration for customers who have an existing outsourced datacenter agreement.
It is doubtful that customers had what Appleby stated in this quote. HANA implementations ended up being quite limited, typically limited to BW, which would usually have three environments. Therefore this is an exaggeration on the part of Appleby. Even in 2019, HANA is still only roughly 4% as popular as Oracle, as estimated by DB-Engines. However, our view is that because DB Engines includes media mentions in its estimation method, HANA is likely less than 2% as popular as Oracle. As evidence, HANA falls off of the lists of other databases entirely. And HANA was even less popular as of 2014 when this article was published.
Appleby’s story here just sounds fishy, and it most likely a lie.
BW is even by 2019 still not in the cloud, and neither is HANA. HANA is still mostly on premises. HANA can be brought up on either AWS, GCP, or Azure, but this is to get HANA more popular. Our case study on HANA on Azure shows problems using it on Azure, as we covered in A HANA Performance and HANA on Azure Case Study.
Referencing Steve Lucas?
Final Words
I’m going to end this blog by stealing Steve’s words from his original blog, because they are still perfect 2 years on.
So back to the original question: Does HANA replace BW? I suppose the answer is: No. But putting a statement out there like “if you haven’t deployed BW you shouldn’t” would be incredibly irresponsible. HANA is definitely many things (A database for BW, a high-performance analytical appliance, a platform for new applications), but matching the entire “system” known as BW point-for-point is a huge project for any company.
BW is here to stay – it’s just changing with the modern HANA based system. And if you’re a customer considering what to do with BW – my advice is simple. Put it on HANA.
Steve Lucas is a highly unreliable source on HANA or databases in general.
Appleby was correct that BW was here to stay, even though using BW undermined the arguments for HANA in a significant way.
What is the Real Number of HANA Customers?
An anonymous commenter stated the following in this article.
“Hi Jon, just to point out that Bill McDermott said that there are 3,200 HANA customers now, of which 1000 are on Suite and 1,200 are ISVs. I don’t know since when companies started counting ISVs on developer licenses as customers.
This means is that there are only 2,000 real HANA customers to date, with the growth trend unclear without more detailed information.”
First, those 1000 customers that were suite on HANA or SOH were terrible decisions for those companies, as we covered in the article Does Moving to ECC on HANA or Suite on HANA Make Any Sense?
Secondly, the commenter is on to something quite true. SAP has since HANA was introduced been exaggerating the number of HANA customers, which we covered in the article Why Did SAP Stop Reporting HANA Numbers After 2015?
This article explains that two quarters after this article was written was the last time SAP reported HANA customer number until writing this article (so March 2019).
Let us see what Appleby says to this.
“So I don’t work for SAP so I can’t speak for Bill. But all software vendors count customer numbers in various ways. You’ll have to confirm what the 1200 ISV number means – but I believe SAP only counts real production licenses.
The real question is the number of live customers and what they are actually doing. Most of my significant HANA customers are doing large-scale BW systems and they are very happy. From a SI perspective, Bluefin’s US and Asia business are now 100% HANA and the UK business also has a significant portion. The revenue growth for us from HANA year on year is significant.
By the way – it’s considered poor etiquette on this site to not post your real name. The assumption is that you have an agenda for hiding behind a nom de plume.”
This is curious. Appleby is “good friends” with Steve Lucas and Vishal Sikka, according to his earlier statement. So it seems as if he could ask them this question to someone other than Bill McDermott. At this time, Appleby was clearly in communication with Hasso Platter, so this is someone else he could have asked. Suddenly, when it comes to answering a question he does not want to answer, he does not seem to know anyone at SAP.
Secondly, the question is not only what customers are doing and whether they are happy. Here it seems it is a pivot from Appleby to get away from the issue. Notice he is moving back to things that cannot be validated — as in “all of his customers are happy.”
Then he calls out the commenter for being anonymous. But Appleby has barely answered the question. To us, this shows fear on his part. Appleby is punishing the commenter while not answering the question. Most of the comments that challenge Appleby are anonymous. This is for a good reason. In the SAP space, you are only able to publish pro-SAP information without potentially facing career consequences.
Conclusion
This article scores a 6 out of 10 for accuracy. But falls when the outcomes of following Appleby’s advice are followed.