SAP has announced a new version of its Business Suite that offers blindingly fast database reporting via HANA with the goal of three-second response times to transactional queries. "It's much better analytics, between ten and ten thousand times faster. I'm not bragging here that we achieved up to 250,000 times faster because …
But is it scalable?
If i was oracle, I wouldn't be picking out my gravestone just yet.
The reason sap got onto hana in a big way is that its database middleware has always been patchy in performance at best - so its solution is to dump everything onto ssd and main memory and access it from there.
Unfortunately this can often lead to an unscalable and unsupportable mess using tools such as the excrible ABAP language and legions of pricey consultants to keep it ticking over on hardware that costs 10 times the price for 2 times the performance.
Time will tell if this leads to real world benefits, but somehow i doubt it!
If i was Oracle I would be trawling my stash of patents...
... looking for something to screw SAP.
Re: If i was Oracle I would be trawling my stash of patents...
with regards to Hana that's a bit hard, since SAP not only offered this technology already several years ago, but Hana was the result of Hasso's University - it started as a research project there, if I remember correctly. Oracle is only making waves about something similar only in the last year or so.
Oracle has too much easy money from their DB software.
To bust a few myths
Hana is not built on ABAP technology.
Hana is fully scalable (IBM have a 100 node test rig with 1TB of memory per node)
Given that it uses relativley inexpensive x86 nodes, prices for hardware will come down
year apon year, unlike Oracle hash job of exadata that still relies on aggregation and extraction
of data and uses propriorty hardware.
Re: To bust a few myths
Agreed, hana is not written in abap but many tools using it will be and no doubt will need " tuning" (recoding) to address performance.
Also, I'm not talking about scalability of hardware, but scalability of performance in the real world with x million rows cubed from sap fi gl posting tables from companies that don't dare to archive... I.e. coalface applications.
All I am trying to say is that performance improvements have been promised before, and in the real world don't match up to expectations especially if you aren't ibm and just don't have the funding for 100 x 1tb nodes...
Yes, prices will come down, but at sap and ibm they may not come down fast enough...
Oracle have made their own bed here, for years they were the de facto database for running SAP on, then they got their own equivalent to SAP and started punting it round more cheaply, or at least stopped the kind of discounts that were offered to SAP customers. Of course SAP are going to cut their Oracle umbilical and release something like this. It's going to be interesting watching this.
I think that the real issue is the historical backups on RMAN, if someone comes up with a method to move these to HANNA so that no Oracle instance is required to restore them, then Oracle will have had it in respect to SAP.
Massive attack on Oracle
DB suppliers like Oracle are going to lose bigtime on this change. They will use aggressive pricing to battle Hana but the big savings to SAP customers is going to a much simple infrastructure, removal of data warehouses in many cases and in the future we will probably see just one SAP system (not seperate ERP, CRM, SRM, BW etc) handling all processes since Hana is fast enough to handle everything in one box.
Re: Massive attack on Oracle
Yeah and sap will sacrifice their licenses for all these systems? I doubt it...
It wasn't that many years ago that we had a relatively small database of projects (a few thousand) and a simple inquiry needed about fifteen minutes to execute. After putting up with this insanity for several years, they finally restructured the hierarchy so that our poor SAP server could trawl though a couple thousands records at a faster rate than four inquiries per hour.
Even now it can take a good hour to perform a business function that could previously have been done and dusted in three minutes by hand.
Perhaps we're doing it wrong. Along with all the other ERP fiascos around the world. It must be us.
Cure worse than the disease. ERP probably singlehandedly explains the endless recession.