In-Memory is the real Winner
In-Memory has been the buzz word in the SAP world for a while and a point of discussion when SAP consultants get together.
While the focus on Big data could have possibly reignited the In-Memory discussion it has definitely got its attention this time.
I want to use this Blog to discuss how In-Memory is influencing the SAP ecosystem and how other database vendors are adapting to these changing times.
To set the context …
SAP HANA is a SAP in-memory Appliance which has the capability to Extract , Model , Analyze and report Massive Volume of data in real time.
When the price of the RAM went down and Processors got a lot cheaper In-memory got its traction. The speed development of the Disk could not keep up with the CPU speed . While traditional databases completely depended on the I/O speed between the disk and database to deal with the performance requirements. With In memory the bottleneck was shifted from disk to the data transfer between main memory and CPU cache.
The entire database could be maintained in the RAM and thereby increasing analytical and transactional processing by several Folds . The effect of this and the gap between OLAP and OLTP closing down Realtime transactional processing has now become a reality.
SAP had timed its HANA release at the right time and disrupted the Enterprise database space, raising the expectations of an ERP Database and the real time Business intelligence derived from it.
Setting the Hype aside , the feedback from its adopters have been pretty impressive…
IBM partnered with SAP on a HANA implementation for Usha International . They went for the IBM Systems solution for SAP HANA based on IBM System x 3690 X5 and x3950 X5 Workload Optimized servers. They experienced a 85 percent reduction in time taken for query execution (approximately seven times faster), with database read time reduced by almost 99 percent. The overall data load time has reduced by 90 percent.
Another impressive success story comes from Medtronics , which is a world leader in medical devices technology and was also a early adopter of HANA. They had a need to analyze large volume of structured and unstructured data from multiple sources to compare customer complaints with sales data . They have reported a Query time drop from 3 Hours to 3 minutes.
While we hear these great stories and successes we will still have to wait and see how SAP vision on HANA materializes. The adoption has been slow and steady and the competition has also realized the importance and significant role that in-memory databases are going to play.
IBM has released DB2 with BLU acceleration which is certified by SAP for Netweaver BW and DB2 near-line storage for SAP NetWeaver BW (SAP Note 1819734 – DB6: Use of BLU Acceleration ). SAP has certified DB2 BLU with several restrictions . SAP hasnt supported HA/DR and Purescale capability as well as paritioning features in DB2 BLU.
Oracle is expected to have in-memory options with its 12c version of the database, there’s already a lot of comparisons between HANA doings its rounds.
SAP support for 12c and the features that will be supported is something that’s going to be closely watched by the vendors . Oracle is currently testing various 12c option and several features for SAP (Ref http://www.oracle.com/us/solutions/sap/sap-database/index.html)
SAP is clearly betting its future on HANA with their entire cloud portfolio powered by HANA .
While every database vendor supporting SAP today already have or working on their in-memory version it will be interesting to see how they position themselves and how SAP handles it.Assuming SAP provides full fledged support to Oracle and IBM versions down the lane, The fact that both these vendors are providing their Inmemory version as a plugin/upgrade to their existing databases may be their selling point along with the whole lot of features that they will offer. Customers however are reaping a lot of benefits using HANA as a appliance in a datawarehousing and reporting scenario where one can expect HANA to continue have its effects.
In-Memory technology seems to be the real winner here and could very well be how ERP databases run in future.