Overall Satisfaction with SAP HANA
SAP HANA had an extended evaluation to become our new relational database to replaced Sybase Adaptive Server Enterprise. Our group reviewed it for two possible use cases. The first which was a step one in our evolution, was to use the A4A option where we migrate all of the data to HANA and use Sybase ASE as a front-end for HANA so that we don't have to recode our suite of applications. ASE would use the A4A feature (Accelerator for ASE) where ASE would push down all queries to HANA to be serviced. This would allow us to get the speed of HANA without the expense of retooling our applications to talk directly to HANA. Then step two was to identify which applications would survive long-term and migrate those to native HANA. We were interested in HANA because of the in-memory performance. During our testing we saw query speed improvements of 10,000X. Not 10,000 percent but 10,000 times faster in HANA than in ASE.
- In-memory performance.
- Seamless integration with Sybase ASE.
- Support for ANSI SQL.
- Support for History Tables.
- Ability to perform on the fly what if calculations with lightning fast speed.
- HANA is not just a database, it's really a platform.
- I wish they would have built-in support for T-SQL since they own that parser from Sybase ASE.
- 10,000 times performance improvement on our transactions which improves our SLAs and response times.
- The multi-node solution improves our uptime and minimizes risk for a component failure.
- HANA allows us to simplify our environment with one copy of the data to service all our our requests instead of 2 copies.
We are able to demonstrate real-time analytics to show how what-if scenarios would impact account balances. For example, if you increased your contributions to your retirement account by 1%, 2%, 3%, 5%, etc, we could show you in real-time how that would impact your balance going forward given certain assumptions (average rate of return, cost of living adjustments, market down years, etc.). We can also model what rebalancing would look like when applied to a large group of accounts and do so quickly. Gone are the days where we have to take a snapshot of our data and send it to the analytics group and wait 2 weeks to have the analysis done. We can self-service and do it real-time before the answer is stale and of no use.
We use mostly relational data, however we are also storing data using JSON documents for noSQL data. We aren't currently using spatial or geo data but would like to apply that to help fight fraud. Our databases have approximately 50,000 data elements that are stored as SQL elements because we had a SQL database. Now with HANA, we can convert some of that data into JSON documents and make our data model simpler and easier to understand.
We rely heavily on stored procedures exactly for that purpose to avoid moving 1,000's or 10,000's rows of data to our middle tier for processing. There are two schools of thought here - to move the logic out of the database to a service. But that doesn't always work from a performance standpoint. By invoking the stored procedure, we can also help defend against SQL injection attacks. Right now we use that for transaction processing and some analytics.
HANA is basically a database platform. I think of it as going full circle back to being mainframe-like. HANA is a platform with everything you need if you chose to deploy it that way. I feel we went to distributed because of the cheap cost but needed many machines to get back to the performance of the mainframe. Now HANA takes us back that way w/o the mainframe costs. But HANA isn't cheap and I don't want to imply that. I'm sure the price may have come down since we did our extended evaluation, but we were looking at a 3-node cluster to handle our data needs. I feel that Exadata offers similar performance capabilities but in my opinion costs even more than HANA. We saw 10,000 times performance improvement on a full rack of Exadata. For HANA, we were running on a much smaller machine and saw the same 10,000 times performance improvement.
Super fast and very easy to use. Also SAP provided us all the support we needed to get up and running. It's an excellent product from a great team. And the A4A option makes it easy to get quickly up and running without the need for rewriting all of the applications.
Do you think SAP HANA Cloud delivers good value for the price?
Are you happy with SAP HANA Cloud's feature set?
Did SAP HANA Cloud live up to sales and marketing promises?
Did implementation of SAP HANA Cloud go as expected?
Would you buy SAP HANA Cloud again?
HANA has incredible query processing speeds. When compared with Sybase ASE for large dataset queries, we saw 10,000 times performance improvement. And for the individual point queries, we found that it was on par with Sybase ASE and that is because all of the response time is the round trip latency from the client initiating the query to the database and back. But for those large set processing requests that take a long time on Sybase, they ran 10,000 times faster on HANA. HANA is expensive so it should be used where in-memory transaction speeds are required and there are large volumes of data to be processed.