TectED over, long live ASE…
Yet another TechED is behind us. Besides the fun of getting reconnected… Besides ambitious meetings fraught with expectations… Besides funny labels and stickers to pass to your colleagues as memento… What actually is there for us ASE professionals in this 20-years celebratory event?
On the one hand there is this: it is always a good thing to get reassured that the product you work with is still relevant. Yes, it is not on the front page with all the buzz that Hana and, now, Vora occupy, but it is still there. It’s being used. It’s being improved. It’s being stuffed with new functionality. It’s being integrated further into Data Management family. It has its own roadmap spanning next 10-15 years. Cool.
On the other hand, what is in it for me – or WiiiFM? If there is one thing I am to take as a give-away from this conference that has practical application for me as ASE professional what would that be? IMO the most intriguing functionality revealed at this TechED has been integration of HANA into ASE as a reporting engine. It is intriguing not only because now it will be possible to solve instant reporting problem by having this stuff delegated to HANA. It is intriguing since in political parlance ASE is at last given legitimacy to remain an OLTP front-end, doing what it is known to do best (or better?) while delegating the things it has to struggle with to a sister engine.
SAP has released express editions to make getting a taste of this functionality first hand possible: ASE Express Edition and HANA Express Edition. Both have limitations.
ASE Express Edition runs on Linux alone (just as HANA is in general – and probably for this same reason). There is a conflicting report which features are open on it – I encourage you to install it and report back to make this aspect clear. But at least it is now armed with 4 engine free license. It is also restricted to 50GB DB size (not sure if this is total data limit for the server of the size limit for each DB it contains – to be clarified as well). Let’s hope Express Edition will be further extended to run on all platforms, otherwise it’s popularity will by very limited.
HANA Express Edition is limited to the minimum memory block it can license – 32 GB. Other than that – it has all its analytical arsenal packed in for free.
So there you have it. A4A – or Accelerator for ASE as the option is labelled – is a first sign that HANA retreats from its stance as all-purpose DBMS and leaves the OLTP (or xOLTP) arena to those who played there well already for the long long time. I hope to see this direction gaining momentum.
Other things I liked to hear:
HADR option is getting more robust, with a possibility to have HADR to HADR replication – topology where both the primary and the replicate nodes are HADRs in themselves. Even more exiting is that stream replication mechanism which HADR is built on is planned to get released to general public next year. My only hope is that this will not be a licensed option (like ASO is) so that general public will be able to exploit its faster performance without having to pay more.
ASE xOLTP functionality remains the same for SP02 with Latch-free B-Trees, Lock-less Buffer Management for Data Caches, NV Cache extension (SSD Based) and Transactional Memory / SNAP (simplified native access plan) on the fitting platforms. SP03, however, will have more fun stuff inside, as it will add In-Memory Row Store with MVCC support to reduce contention even further. The idea behind is that hot data will be elevated into an in-memory row store where locking is lifted by versioning.
All in all the event was fun… even for me…
ATM.
What exactly does
"integration of HANA into ASE as a reporting engine"
mean in practice ?
Is there anything on the roadmap for ASE other than xOLTP ?
You said
"It's being stuffed with new functionality."
What functionality is being added ? Is it under the covers changes to improve OLTP performance or will we see new features ?
I guess what I'm asking - what industry is being targetted ?
Mike,
Integrating Hana into ASE means you may attach Hana to ASE as a replicate side and pull data from ASE into Hana using rep server. When you attach the replicate side it comes as Hana on back-end + ASE on front-end. Thus your reporting replicate side is visible to the application as just another ASE (therefore it's called Hana + Adapter 4 ASE). When you have your data in Hana you may then query it through that front-end ASE. ASE is "taught" to push most of the optimization your query needs into Hana so that it is performed mostly there giving you the response time Hana is known for. This new architecture allows you to have your legacy reports run without any code change on your "reporting ASE" block which is in fact Hana engine dressed up as ASE. You gain speed without bothering to rewrite your code. You also gain access to 100s of other goodies wrapped into Hana engine. You can, say, then throw your ETL code and base your BI on Hana (instead of pulling it, e.g., into MSSQL server nightly). You are given the capability - what you do with it or if you find it worth messing with is completely up to your imagination.
"New functionality" mostly targets xOLTP. It does not target any particular industry. What it does target is removing contention on as many levels as possible as the processing speed in all the industries generally speeds up. Lot's of traditional customers (banking? insurance? retail?) move more and more into SSD space. With this come new challenges and the need to clean up contention (locks, latches, spinlocks &c) in order to make faster processing possible. It's called xOLTP (and unfortunately streamed through MEMSCALE option) but it basically aims solving contention in the age that has faster IO response time. There's more to it (like, say, dynamic tiering within ASE itself) - but the aim is basically the same: faster response time, lower contention.
I hope I've answered your questions - at least partially 🙂 .
Cheers,
Andrew
Andrew,
The ability to connect HANA to ASE and be able use ASE T-SQL to query HANA is quite neat. Will be interesting to see how it works in practice - will cross DB joins allowed between ASE and HANA. ? (and how much the extra hop works in practice). If I have to copy all Tb's of data to another system it could be expensive.
Thanks for the information on "New functionality" - I wouldn't term moving to xOLTP "New Functionality" which I guess thats' why you've put it in quotes 😉
Personally, working in finance we're not interested in xOLTP - we're looking for features like temporality, more datatypes, better T-SQL, etc
Mike,
If an ASE customer has a secondary "reporting server" ASE, s/he can use HANA as the reporting engine under the covers, speeding up the reporting by 100x in some cases. This is being termed as "SAP HANA Accelerator for SAP ASE". Please see more here: https://help.sap.com/hana_options_ets