Analysis of Steve Lucas’ Article on What Oracle Won’t Tell You About HANA

Executive Summary

  • Steve Lucas is one of the most quoteworthy of the SAP executives. In an article, he proposed many false items around HANA under the concept of “setting the record straight.”
  • Steve Lucas discusses topics ranging from performance to pricing. In this article, we review Steve Lucas’s level of accuracy.

Introduction

Steve Lucas wrote a famous article back in 2012 when HANA was first introduced. Now five years later, we will review what came true.

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.

  • This is published by a research entity, not some lowbrow entity that is part of the SAP ecosystem. 
  • Second, no one paid for this article to be written, and it is not pretending to inform you while being rigged to sell you software or consulting services. Unlike nearly every other article you will find from Google on this topic, it has had no input from any company's marketing or sales department. As you are reading this article, consider how rare this is. The vast majority of information on the Internet on SAP is provided by SAP, which is filled with false claims and sleazy consulting companies and SAP consultants who will tell any lie for personal benefit. Furthermore, SAP pays off all IT analysts -- who have the same concern for accuracy as SAP. Not one of these entities will disclose their pro-SAP financial bias to their readers. 

Steve Lucas Quotes

“The recent and significant increase in negative messaging Oracle is attempting to spread about SAP HANA is pretty amazing. Traditionally at SAP, we’ve taken the high road when it comes to responding to these claims, as nearly 100% of what Oracle says is inaccurate and designed with one thing in mind: To protect their established revenue base. All you have to do is look at what Oracle has stated about the cloud over the last 10 years in order to understand their plan of attack for in memory:”

  1. Step 1: Halfheartedly acknowledge its presence
  2. Step 2: Continue to push old technology
  3. Step 3: Create fear and doubt related to innovations

“We see the same, tired response plan from Oracle emerging for in memory and SAP HANA.”

Steve Lucas works for SAP and is not an unbiased source of information on SAP. So these claims against Oracle are also designed to improve revenue.

However, are they true?

SAP Taking the High Road?

Since I began working with SAP in 1997, I have never seen SAP take the high road. SAP is a consistently unethical and monopolist software vendor, just like Oracle. Secondly, Steve Lucas has not worked with SAP all that long. He came over from Marketo, so he would know less about SAP’s history than I would. Thirdly, through his comments, Steve Lucas demonstrates that he knows very little about databases.

Therefore, he is unqualified to either promote or speak on SAP’s database or contradict Oracle’s claims. Oracle also makes exaggerated claims about its database(s). But one has to know something about databases to oppose them or identify which claims are the most sensational.

SAP Using a Similar Storyline to Oracle Regarding the Cloud

Interestingly, the items pointed out by Steve Lucas regarding Oracle’s response to the cloud have also been SAP’s response to the cloud. That falls under the category of hypocrisy to the highest degree. Both SAP and Oracle are on-premises vendors that have tried to put off the cloud as long as possible while cloud washing.

“Why would Oracle want to innovate anyway? There are billions of dollars’ worth of reasons why Oracle must protect their legacy database technology. Oracle has boxed itself into a corner, where they can’t afford to cannibalize their existing revenue stream, and at the same time are obligated to push massive amounts of hardware from their Sun acquisition – effectively amounting to a digital albatross hanging around their neck.”

I could make and have made the same argument regarding SAP! SAP is not an innovative software vendor but poses as one. I have proposed that SAP reduces innovation in the enterprise software market in the article How SAP Reduces Innovation.

Billions of Dollars of Revenue to Protect

Likewise, SAP has billions of dollars’ worth of reasons to protect its ERP and other application business by raising as many barriers as possible. By paying off IT media sources by giving their implementation business to unethical consulting companies like Deloitte, Accenture, and IBM. These then recommend SAP regardless of the fit between the SAP application and the business requirement by creating a false construct of Type 2 indirect access to block out competitors who have beaten them in software selections.

So SAP has no stones to through here, and if Steve Lucas can’t figure that out, well, I don’t know what to say.

“We should also let it be known that comparing this Exalytics bundle/packaged thing is just a red herring for customers. It’s an attempt to imply that HANA isn’t ready to compete with Exadata or Oracle’s core database. It’s an attempt to imply HANA is limited to analytic business scenarios.”

HANA is limited to analytic business scenarios. Steve Lucas should probably have educated himself on databases before writing these things. HANA has portions of the database that is column oriented. SAP proposed that the entire database was column oriented, but we later learned that HANA is a combination of row and column-oriented tables. However, HANA is useful only for queries run on preconfigured analytic data (simple SQL queries) and slow for transaction processing or complex or lengthy queries.

Column-Oriented for Analytics

It is well known that column-oriented databases are optimized for analytics. This is covered in the article, Why SAP HANA is So Fast. Oracle 12c has a superior ability to switch between column-oriented and row-oriented storage. This is included in the article, Which is Faster HANA or Oracle 12c?

Steve Lucas’ writing is like reading someone falling down a set of stairs. Steve Lucas seems not to understand databases but is very comfortable discussing them.

“Bottom line for Oracle, there is no incentive to truly innovate. Is innovation loosely coupling and stacking an RDBMS + TimesTen + Exadata + Endeca on expensive hardware? Clearly they would rather repackage 20 year old, tired products on refrigerator sized servers and overcharge customers for it.”

What a statement. Oracle did innovate with Oracle 12c several years after this. Oracle has been improving its database for decades since it was first introduced. It is nothing compared to how much money Oracle brings in from their database, but they improve the database. Moreover, why is Steve Lucas obsessed with Exadata in this article. Did he not comprehend that Oracle was pushing its analytics appliance as HANA is primarily an analytics database!

Is that point lost on Steve Lucas?

“In this blog, I will provide a set of facts about SAP HANA and in-memory computing. I will try my best to be objective throughout and dispel some of the silliness Oracle is spreading.”

In Memory Computing?

In memory, computing is a myth, which is something we covered in this article.

In memory, computing was aggressively promoted by SAP, and instead of explaining its inaccuracy, Oracle decided to copy SAP’s marketing message.

Interestingly, Steve Lucas lacks the self-awareness to describe silliness written by others when he does not himself have the subject matter expertise to understand this topic.

“Let’s take a look at the inaccurate comparisons and misnomers floating around:”

  1. “TimesTen is a mature database technology vs. SAP HANA.
  2. Scenario comparisons of Exalytics vs. SAP HANA
  3. Pricing comparisons between HANA and Exalytics.”

Is misnomer the right word here? A misnomer is a misnaming. It is difficult to see what misnomer Steve Lucas is referring to. Steve Lucas means misimpression.

Here are his explanations for why, in essence, SAP is wrong.

#1: Comparing Database Feature/Functionality:

“Oracle is attempting to compare TimesTen to SAP’s HANA database. There are as many 14 attributes that they claim HANA lacks, including in-memory aggregates, multi-dimensional OLAP (MOLAP), in-memory indexes, NUMA support, etc. Ironically, most of these are cumbersome mechanisms that Oracle has patched on to their RDBMS to improve performance, but thanks to HANA’s innovation, are unnecessary overhead and maintenance tasks that customers will be glad to leave behind.”

“To me, this comparison at a very basic level is exactly the same I would have expected to hear from the horse and carriage salesmen when automobiles were first introduced. Oracle is essentially arguing that our “car” is not as good because it doesn’t come with feed for the horses, or a large bucket and shovel that customers can use to clean up their mess.”

This is highly misleading.

And it is not Oracle that has “patched” them onto their RDBMS. They are part of any RDBMS. Steve Lucas may be surprised to learn that aggregates are a pre-computation device used to speed processing. If a series of values are consistently referred to, it makes sense to create an aggregate table. SAP seems to see this as some sin, but it is a sound practice. MySQL has aggregates, MariaDB has aggregates, you get the idea.

H0w Steve Channelled Hasso Plattner

It is unclear why they are considered cumbersome by Steve Lucas, but I suspect this is only because Hasso Plattner told him this.

Hasso Plattner places a lot of emphasis on removing aggregates in this book. Hasso wrote four books on in-memory computing, all of them filled with falsehoods. However, outside of SAP, others that focus on databases and have concentrated on databases far longer than SAP do not agree that aggregates are a terrible thing. We are currently developing an application, and one has to decide whether to precalculate or whether it is worthwhile to re-calculate values. Hasso Plattner seems to emphasize reducing aggregates because he needs to show that he and, by extension, HANA made essential contributions “innovations” to the column-oriented design, when in fact, he didn’t. We cover Hasso Plattner’s lack of innovation and the fake storyline around HANA’s innovation in this article. 

Also, aggregates aren’t maintenance items. They mostly sit there until called upon. Indexes are very similar to aggregates but are pointers for the database to find things more quickly. Indexes also make a lot of sense.

The Major Problems with Steve Lucas’ Comments

So there are three major problems with Steve Lucas’ explanation here of RDBMS systems.

  1. It proposes that Oracle is the only database vendor that uses indexes, MOLAP, aggregates, etc..
  2. It proposes that all of these items were grafted onto the RDBMS willy-nilly. They weren’t; they are part of the design.
  3. It proposes that these additions are a lot of maintenance when they aren’t.

There is some truth to the carriage versus the automobile, but the problem is it shows Oracle as the aggressive party here when SAP, that is, the party stating that they have come up with some quantum leap in databases when they haven’t. SAP is saying that all of these items are irrelevant when they aren’t. For instance, HANA still uses aggregates, although it minimizes them.

There are, in fact, cases where it can make sense to replace Oracle, but the opportunity is not with HANA. The chance to replace Oracle is with lower cost and quite competitive databases like PostgreSQL or MariaDB and moving Oracle DBs off on premises to AWS or Azure that allow for those databases to be managed. Once the database is managed, the reason for having Oracle declines. Oracle has its place, but it is overused because it is considered the standard, and IT departments are comfortable. But most Oracle databases are used far below Oracle’s capabilities. And databases like PostgreSQL or MariaDB, or AuroraDB now cover the vast majority of most company’s needs.

Steve Lucas’ Facts?

Here Steve Lucas presents what he calls facts. Let us review his list.

A few facts:

HANA is an ACID Database!

“SAP HANA is a fully ACID compliant database”

All the databases that HANA is competing with are ACID. This is like saying that a car competing with another car has “wheels.”

HANA is Entirely Loaded into RAM?

“HANA’s design manages and accesses data completely in RAM allowing for speedy retrieval of data over massive volumes and addressing the BigData problems of today and the future.”

False. This was HANA’s original presentation, but the data is swapped in and out of RAM with HANA. It is just a misrepresentation of the database.

HANA Does Away with MOLAP?

“HANA does away with the need for MOLAP, tuning structures such as multiple indexes, aggregates, materialized views, and saves the costly time to build and maintain such structures.”

Yes, that is true. That is the benefit of the column-oriented design. But IBM, Oracle, and Microsft also now have this same capability.

HANA has Super Efficient Queries?

“HANA leverages parallel queries to efficiently scale out across server nodes as proven in our April 10 scalability testing announcement in which 100 TB of data partitioned across 16 nodes was queried with sub-second response time.”

First, Steve Lucas has no idea what this sentence means. But if we evaluate the sentence on its merits and leave Steve’s knowledge out of it, if this were true, it’s not showing in our analysis of both public and private benchmarks. SAP is not the performance leader in any category versus the competing offerings.

HANA is a Good Place for Unstructured Data?

“HANA handles both unstructured and structured data and has since its inception.”

All databases can do this. The question is whether the database is the right one to use. Due to HANA’s high cost per GB, it will not be the right DB to store unstructured data. That is why SAP introduced Vora to connect HANA to Hadoop. But Vora is dead in the market.

HANA Can Store Non-SAP Data?

“HANA handles both SAP and heterogeneous data equally well.”

Well, data is data. The database is not going to care what application the data originated from. But HANA is not used to store non-SAP data because HANA is only sold into SAP environments.

Oracle Can’t Scale Beyond One TB

“Oracle has not demonstrated how Exalytics with TimesTen can scale out beyond the 1TB limit and Oracle has publicly stated that the usable memory in this configuration is about 300GB.”

Not sure what to say to that.

HANA is Super Innovative?

“HANA does, in fact, support ANSI standard SQL syntax, as well as MDX. Just like Oracle extended the ANSI standard with their PL/SQL procedural language, SAP has extended the ANSI standard support in HANA with SQLScript, a procedural language that allows you to write programs with logic not possible to implement in the single-statement SQL language.”

“HANA allows you to manage your data as you choose in either column stores, row stores, or a combination of the two. (Plus other stores/models) Oracle’s claim that you must load data into the row store, and then migrate to the column store, is false. Not to mention their claim that columnar data must be migrated back to the row store to be updated and then back to the column store to be queried. This is simply not how HANA works at all.”

The bottom line: HANA is a next generation solution that replaces many of the tired, legacy products that Oracle continues to re-label as “innovative”.

None of HANA’s claims related to innovation regarding HANA have held up. It is improbable that SAP, with its far lower database expertise, came up with a better way to manage row and column data than Oracle. But the point is moot because performing OLAP and OLTP from the database is both not a critical requirement for companies, and no database can perform OLAP and OLTP optimally for both.

Our Conclusion on Section 1

One can go back and forth on each of these bullet points. Some are false. For instance, HANA is not designed for unstructured data and is not priced for holding Big Data. HANA is not a player in the Big Data market. HANA is priced per GB and is very expensive per GB, so even if HANA is good for unstructured data, it is too costly to place much data in HANA.

I cannot see why Oracle’s database is “tired and not innovative.” Part of what is happening is that Steve Lucas is commingling being immature with being innovative. Yes, HANA has more releases, but that is not necessarily a good thing. It means less stability, and HANA can’t compete with Oracle 12c in how SAP promised that it could. Oracle 12c has a far better reputation for reliability than does HANA and for fault tolerance.

#2: Comparison of Use Case Scenarios Between HANA and Exalytics

“The easiest way to find out how HANA is transforming our customers is by looking at the customers stories on saphana.com. Anyone can publicly see proof – compelling business case after case where customers are consolidating IT systems and delivering breakthrough business value with a lower TCO than Oracle.”

SAP would not have jerry-rigged these case studies, would they? For instance, would SAP place a case study on their website where the customer had many problems with HANA or where it failed to meet expectations? I say this because I know of several instances where HANA was unable to meet different customers’ expectations, but this did not seem to make it to the SAP website.

SAP’s History with TCO Claims

Second, SAP never estimates TCO. TCO calculators of various SAP applications and non-SAP applications are available for free at the following link. In each of the application categories where SAP had an application, SAP came out as the highest TCO. SAP’s high TCO is why they have so many consulting partners. Deloitte, Accenture, and IBM, and others build their consulting practices around SAP for precisely why they can make the most money on SAP implementations. SAP’s proposal about lower TCO is entirely made up, and in fact, any statement by SAP about TCO can be dismissed out of hand because SAP has lied so frequently about their applications lowering TCO.

“I have a moral objection to this whole comparison Oracle attempts, as the mere existence of Exalytics is just a diversionary tactic. Oracle doesn’t want HANA digging into their RDBMS business or Exadata for that matter, ergo the attempt I mentioned at the beginning of this blog.”

If Steve Lucas had any morals, he would not be a top executive at SAP. Also, for a person with such high moral standards, he lies a lot when he writes. Therefore, this statement regarding any moral concern must be false. And secondly, Oracle’s comparison between Exalytics and HANA is quite logical because HANA is a specialized analytics database. It isn’t easy to see why Steve Lucas can’t see this. But Steve Lucas might want to ask people at SAP that know about databases why SAP stopped releasing a transaction processing benchmark for HANA.

Nothing just written should be taken as an endorsement for Oracle’s morality; it is well known it has none.

Steve on What HANA Supports

“Let me state (again) that SAP HANA supports analytical functions (e.g. all datamarts – e.g. T-Mobile’s use of customer micro segmentation analysis across millions of customers, across SAP and non-SAP application systems), business functions, planning functions and predictive functions (SAP BusinessObjects Predictive Analysis and Predictive Analytical Library in HANA) as well as transactions natively (e.g. upcoming SAP ERP on HANA). To do this in Exalytics you need – TimesTen, Essbase, Endeca, Oracle RDBMS, etc. This amounts to more money for Oracle and no business breakthrough for the customer…not exactly a “win-win”.”

Support is a different word than “excels at.”

Any database can support virtually any application. But the question is whether it does it well. It turns out that HANA is a poor match for S/4HANA, but this not yet a problem because five years after this article was written, S/4HANA is still not ready to implement.

This falls in the conventional approach of merely accepting because SAP offers something or supports something that must be good or worth using. What other vendor gets that kind of slack cut for it?

Steve on What Scenarios HANA Can Address

“What Oracle doesn’t tell you about SAP HANA is that it addresses scenarios that were not feasible before, even with an armada of legacy tools – for example with the business function library in HANA it is now possible, using standard SQL, to execute in-database processes and functions that could not be written in SQL before.”

Steve Lucas has swung and missed on far more straightforward assertions already in this article. How would he possibly have the background to know if this is true. And this may or may not have been right in 2012, but Oracle 12c has things that Oracle 12 did not have.

SAP BW Will Not Need to be Recoded for HANA?

“With respect to SAP BW customers, Oracle has argued that customers would have to re-code BW applications to work on HANA.This is simply not true, and we’ve published numerous public statements from our customers that have already migrated to BW running on HANA in production.”

No, it is true. And here is why.

BW is based upon a relational design, but it uses emulated cubes based upon a star schema. BW is one of the least productive report data structure applications that I have used. There are plenty of bad ones that come to mind, such as IBM’s near dead Cognos. However, BW absorbs massive amounts of human resources and produces very few reports. A lot of what BW does is take flat files and make them into a star scheme called InfoCube in BW. Then queries are run on the InfoCubes. Star schemes, or as SAP calls them in BW, InfoCubes speed read processing. These InfoCubes are rendered unnecessary when a column-oriented database like HANA is used along with BW because the column-oriented design and the memory significantly speed the read. However, the problem comes in that the customer has already spent vast amounts of money building all of these InfoCubes. InfoCubes that are no longer needed. So what is to be done?

Justifying the Investment into HANA

To explain the investment into HANA, but not undermine the investment into the InfoCubes, most customers that put BW onto HANA keep using the sale InfoCubes. But when new aggregations are necessary, the BW resources run merely the query against a non-cube based structure called an InfoProvider. Essentially the BW is in contradiction with HANA because the BW was created to run on a relational, or more accurately, a row-oriented database that was not optimized for analytics.

Therefore, one can dispense with recoding BW after moving to HANA, but there is a natural conflict between BW and HANA. SAP may have published numerous public statements on the topic, but the problem is that most of SAP’s public comments tend to be either false or misleading. The fundamental conflict between BW and HANA will, of course, not be addressed by SAP.

Our Conclusion on Section 2

Steve Lucas is attacking the reader with an unending list of false contentions. The only things that Steve is pointing to that are right are also true of the databases for which HANA competes.

#3 – Pricing Comparisons

“Oracle has tried to publicly compare Exalytics pricing to HANA and the information presented is grossly misleading. Not only is SAP HANA less expensive in up-front cost than Oracle’s Exadata + Exalytics bundle, (plus all the derivative components you’ll need to make it work) the differences in total cost of ownership are substantial.”

The TCO of HANA and Exadata is Known?

Steve Lucas has no idea what the TCO of HANA or the TCO of Exadata is. Since SAP was established until this article was written, I have never seen a single credible publication on TCO by SAP.

HANA’s Pricing?

“We didn’t just innovate on the technology platform with SAP HANA, we innovated on pricing as well with a straightforward, simple to understand model based solely on the amount of data held in memory (Unlike Oracle which charges per CPU plus test and development environment fees.) A single unit of HANA (1 HANA unit = 64 GB of RAM) includes the FULL production, test and development licenses a customer needs. It also includes the data modeling and management tools needed to get data into HANA and actually use the product. Even better, HANA gets cheaper over time…the more you buy, the more the list price per unit reduces.”

Putting Business One on HANA?

Wow. This is a lie. HANA is exorbitantly priced. And SAP has not been transparent about the pricing but has instead been publishing information meant to give prospects a misimpression of the amount HANA can be reduced in its footprint. SAP has been utterly opaque about HANA costs, as they have endeavored to hide HANA’s real costs from customers.

“Our customers running SAP Business One can purchase a HANA license for as little as €2K. Any customer can purchase a license of SAP HANA Edge Edition for €40k. We also have the SAP HANA Netweaver BW edition for as little as €13k per HANA unit. (64GB of RAM = 1 unit)”

HANA’s Hardware is Not Expensive?

Why would any customer put BusinessOne on HANA? BusinessOne is targeted towards smaller companies that purchase BusinessOne because it is relatively inexpensive. Why would a price sensitive company like this put BusinessOne on HANA, a database that is both expensive to buy and very expensive to maintain? Even the pricing Steve Lucas is putting forward here does not match the pricing list from SAP in any way.

“I am certain Oracle would cite that hardware is extra, so let’s cover it right now. HANA servers from certified partners like Fujitsu are available for as little as $12K. That’s because we don’t force our customers into one hardware stack and then overcharge for it. We have certified partners like IBM, HP, DELL and the list goes on and on for HANA. We believe that the Intel platform, combined with the continued commoditization cycle we’ve seen for the last 40 years in computing will win…period.”

Even if the hardware were free, it would not make up for the other costs associated with HANA. I was unaware that SAP hardware is somehow more cost-effective than hardware purchased for other applications.

“Even with rapid growth of data, 95% of enterprises use between 0.5TB – 40 TB of data today. For this market, at the low end (0.5 TB) the combined cost of hardware and software is approximately $500K and at the high end the pricing is comparable to Exalytics alone today. In a recent test, we ran a 16 node cluster of 100 TB of uncompressed data, read more in the SAP HANA Performance Whitepaper.”

I think Steve Lucas should have provided more detail. There are a lot of details about this. Exalytics is expensive, but HANA seems to work out as the most expensive database available. But HANA’s features and performance don’t match up very well to the databases that SAP compares.

SAP Will Win?

“As I said when I started this blog, SAP doesn’t usually comment on competitor FUD, but I wanted to pause and be clear. SAP will win this market because we rely on fact and figures, real performance and customer success.”

So far, SAP has not won in the database market. SAP has picked up market share, but it dramatically lags Oracle12c 5 years later. Of course, it is not clear what time horizon Steve Lucas was forecasting. According to DB Engine, a website that tracks various databases’ popularity based on things like media mention, support tickets, etc..HANA is only the 19th most popular database in current usage. This is right around MariaDB and Hive.

Furthermore, looking at the popularity value, Oracle is 28 times more popular than HANA. Moreover, DB Engine is most likely overestimating HANA’s actual usage because it measures things like media mentions, and SAP has an enormous ability to get media attention on HANA. For example, one hears less about applications like MariaDB that are far more widely used than HANA (and also free, by the way)

SAP Delivers Innovation?

“We are going to continue to deliver real innovation and let customers decide who can better help them build the future, not continue to invest in re-packaging the past!”

SAP does not deliver innovation. They are a marketing company filled with blowhards like Steve Lucas that will say anything to continue to collect millions of dollars of stock options. This innovation narrative is becoming extremely old, and it overshadows real innovation, which happens at other vendors, notably smaller vendors. At Brightwork, we have given only four applications, perfect scores for innovation, and none of them were large vendors.

Our Conclusion on Section 3

The same as our conclusion in Sections 1 & 2.

Overall Conclusion

Steve Lucas is an untrustworthy source of information on HANA. He was certainly not the right person to write this article, and it was one of the lowest quality and least honest articles I have read in a very long time.