Before I left on vacation (now almost a month ago – can’t remember when I had such a long vacation if I ever had), Mark Fontecchio organized a short video conference between myself and John Appleby. The idea was to compare Oracle Exadata with SAP HANA in a shot video discussion. Unfortunately, video part didn’t really work but we did end up with at least a podcast — thanks Mark for organizing it.
Download MP3 — Oracle Exadata vs. SAP HANA
I must say I don’t have any significant knowledge about SAP HANA so my conclusions are based on SAP HANA public materials, architecture overviews, marketing info and my vision of current use cases and trends in data processing and storage space. To my surprise, John and I seemed to agree that HANA and Exadata are currently solving different business problems and you can’t really say they are competing. However, based on SAP direction, they intend to make HANA competing with traditional relational database systems used for SAP products including Exadata. This does seem to me to be couple years away, though.
My limited perspective tells me that momentum for real-time analytics on Big Data has just started to pick up and by the time it’s in-demand en-mass, Oracle would have an integrates solution so customers relying on Oracle software would naturally gravitate to Oracle. John’s argument is that customers could just as well gravitate to their application vendor instead and those using SAP application products would tend to look to SAP HANA more. We both might be right in the end.
Excellent post Alex,
SAP HANA should only be compared with Oracle Exalytics, Oracle Exadata is far far away from the reach of SAP Hana. There are so many problems with SAP Hana that comparing it with exadata cannot be justified. SAP Hana is not a new product is a modification of SAP BWA. which SAP has been working since 2001 but they dont sell.
I think Oracle Exalytics with Oracle database and OBIEE 11g will easily beat SAP Hana. SAP hana uses business object as reporting tools which they have acquired many years ago and there is a lot overlapping in this technologies. compared to it OBIEE 11g is extremely cool.
SAP is announcing SAP BW integration with SAP HANA from 4th of November, Means that SAP HANA will be able to host both real time analytic and tradition analytic. We already have 90% of market in China, India, Pakistan, Australia and middle east working on SAP HANA. SAP claims that their solution will easily beat Oracle Exadata. After this move its perfect legal to compare SAP HANA and oracle Exadata. Most of people on this site are Oracle DBAs and have no idea of data warehousing and business intelligence. SAP HANA is the next force in data warehousing. Exadata is too expensive, complex and ineffective because its been designed for OLTP. Further Oracle Team and Oracle People have no idea what is data warehousing and they consider anything which Oracle tells them as right. In fact, I have seen some big professionals spending their time in doing block dumps to find out how Oracle work. Its totally stupid and non Professional. Oracle is still not able to figure out what to do and they are releasing things like Oracle Exalytic etc etc.
As a person who is working with both Oracle and SAP software I want to throw some stuff here. Oracle Exalytics is purely targeted for BI. Its combination of ESSBASE and times ten (not a pure in memory database, its hybrid).On the other hand SAP HANA is tabular complete In-memory database, it does not store data in cubes, all its performance gains are through inmemory,column data stores and compression and other technologies. SAP HANA is suited for both OLTP and OLAP, as of now SAP is pushing for OLAP(DW) only because security and data safety features are not developed enough for OLTP. But SAP HANA’s performance is so good, that it beats any available cubic and DW targeted database with Star or Snow flake schema ( HANA beats these databases performance getting data from 3NF database). So virtually all ETL, flattening of data, cube loading will become irrelevant for HANA. Just sync 3NF data from ERP to HANA, and run query against it, it runs same or better than getting data from pre-calculated cube.
Coming to the reporting tools I work with OBIEE and BO, frankly speaking, though there were some innovations in 11g, it will take years for Oracle to match the entire feature set available in Business Objects. Future of HANA is (by 2015) it will add Security and recovery features of SYBASE ASE and column store features of Sybase IQ, to become a single database solution for all the requirements. But still we cannot know if all this can happen, but it’s what SAP is saying.
Oracle just ignored HANA initially, after seeing it gaining traction, they made a combination of Essbase and Times ten (which are quite old technologies) to show that there is something comparable.
I worked with Oracle,BO,OBIEE for majority of my carrier, just from few months i am working on HANA,its performance just blows me away. Just from technology perspective, if oracle cannot innovate fast enough HANA will throw it under bus, But in this world not exactly better performing products succeed, it depends marketing and other stuff also, If you talk only about technology.. Oracle is not even close to releasing something of same technical capability.
HANA OLTP performance is not impressive, but may be good enough for many. If it were truly good performance then SAP would publish an SD benchmark, which they refuse to. Also they wouldn’t need to bundle Sybase ASE as part of a “HANA” package for high volume OLTP.
What a surprise! ;)
Do you have hard numbers / facts to refer to?
Here are some links…
Refusal to publish SD benchmark: https://www.asugnews.com/article/whats-the-sd-benchmark-for-erp-apps-running-on-sap-hana
It’s a good read, and contains the following SAP quote:
“We are working on a new set of benchmarks, which is more suitable for Suite on HANA,” SAP says, “but this will not be ready before Q3/2013.”
Got to love that, and as I commented on that same page the new benchmark mysteriously never surfaced.
SAP SD benchmarks: https://global.sap.com/campaigns/benchmark/index.epx
OK, HANA database doesn’t support a 2 tier configuration, but you won’t find any HANA reference in the 3 tier results either.
SAP’s bundling of Sybase ASE as part of “SAP HANA” for “extreme OLTP”:
They really go to extremes to give the impression this is HANA database, but it isn’t. If you hunt around long enough on this you’ll get the true picture.
See the SD 2 tier results for the “extreme” OLTP performance SAP boasts about with Sybase ASE. Despite double the memory and double the processor power (60 x 2.8Ghz cores versus 24 x 3.52GHz cores) Sybase ASE only beats the most recent DB2 10.5 result by 14%.
My take: Sybase ASE is *not* that fast for OLTP, just much faster than HANA. That’s based on SAP’s total lack of willingness to provide SD benchmark data for HANA, and its bundling on Sybase ASE with HANA as a stop gap for customers high transaction rates.
I don’t see scale out as improving Suite on HANA performance easily. There are reasons that other database companies have tended to avoid shared nothing parallelism for OLTP workloads. In particular, two phased commit is a costly extra overhead for short transactions. Maybe SAP can scaleout Suite on HANA through very careful partitioning choice to ensure collocation of related data, but that’ll take quite a bit of application redesign. Interesting times… but meanwhile I’m sticking to my view that the HANA technology lags the hype by a large distance.
Thanks for your comments Paul.
It’s mid 2015, and still nothing about SAP Hana on “SAP SD Standard Application Benchmark Results”.
Thanks so much for this article! I tried to follow some instructions from few other article and got in over my head. This worked so quickly and your instructions were very easy to follow. Really appreciate this.
I’m absolutely new to SAP and am trying to implement the Audit Log on HANA.
I have successfully made it so that every Create, Update, Delete of a record inserts a row into AUDIT_LOG table.
However, many of the values are stored as “”?”” such as PREV_VALUE, VALUE, KEY, etc.
Under the STATEMENT Column, the update statement also doesn’t provide sufficient information.
update “”TDH””.””TABLEEXAMPLE”” PERM set “”SNO””=TEMP.””SNO””,””EXAMPLE1″”=TEMP.””EXAMPLE2″”,””EXAMPLE4″”=TEMP.””ACCOM_PAID_IND””,””VALID_FROM””=TEMP.””VALID_FROM””,””VALID_TO””=TEMP.””VALID_TO””,””ACTIVE_FLAG””=TEMP.””ACTIVE_FLAG””,””LAST_UPDATE_DTE_TIME””=TEMP.””LAST_UPDATE_DTE_TIME””,””LAST_UPDATE_BY””=TEMP.””LAST_UPDATE_BY”” from “”TDH””.””TABLEEXAMPLE”” PERM, “”#EXAMPLE3″” TEMP where PERM.””KEY””=?,(4)
I have edited some information on the statement to ensure our table names are not shown.
May I know what could have went wrong or is this expected?
Excellent tutorials – very easy to understand with all the details. I hope you will continue to provide more such tutorials.