Ingres and JasperSoft, two of the heavyweights of the Open Source community, have announced today that they have certified the latest release of each others software on their platforms. As part of this announcement, they are claiming that JasperSoft’s BI Suite on Ingres VectorWise analytic database delivers faster query response times than any other solution currently on the market.
By throwing out the faster query response times statement, Ingres is relying on its VectorWise analytic database engine being able to deliver on claims from June 2010. At that time, Ingres announced that VectorWise was the first purpose built analytic database engine that was capable of delivering very high speed performance on commodity hardware.
The key to the Ingres claim was that other vendors database products were not capable of taking advantage of features in modern server processors such as multi-threading, SSE and the increase in size of L2/L3 caches. For VectorWise to take advantage, it meant architecting a new database engine from the ground up, something that the mainstream database vendors have not done.
Using a dual socket Intel Xeon x5560 2.8GHz computer with 48GB RAM, Ingres claimed that VectorWise could improve a typical query from 16.5 seconds (based on Ingres 9.0) to 0.206 seconds. That time compares with a hand-written and optimised C++ program to do the same query which took 0.04 seconds. The query itself used a datacentre generated in accordance with the TPC-H benchmark although this was not a formal TPC-H standards test.
What gives VectorWise its biggest performance increase is that it treats the L2/L3 cache on the processor as its main memory and then uses RAM and disk as secondary sources of memory. This in-memory cache approach is something that Oracle, Sybase, IBM, Microsoft and others have been using but with less dramatic effect.
By combining that level of performance with JasperSoft’s BI Suite, the two companies are now signalling that Open Source is not just about niche solutions or for companies on small budgets. This level of performance is something that large enterprises with complex BI solutions are struggling to achieve with their existing solutions that are expensive to run and tune. At time of writing, there were no prices available for an Ingres/JasperSoft solution.
In another challenge to the established database vendors, both companies are stressing that their solutions are standards driven and not just for on-premise applications but can be delivered through multi-tenant SaaS and Cloud platforms.
There are a lot of smaller Cloud solutions providers looking for products that would enable them to not just support the SME but also go after the very lucrative enterprise market. An Open Source pairing that could be quickly dropped onto their servers and then delivered to customers without the complexity of licensing that accompanies the solutions from the big database vendors.