Many of you have reached out to find out how we accomplished such stunning performance at the Strata Big Data Conference last week. The below video talks about our 10X10X10 challenge and provides details around the performance.
Analyzing 10 Terabytes of Data on a $10k server in less than 10 seconds is indeed an incredible performance. Trying to replicate this using other approaches could end up costing 50 times that, either in hardware, software or people! (there is a reason we call this a revolution!)
The innovation comes from two areas – software and hardware. On the software side, our columnar database technology allows us to store and compress data so we can efficiently make use of machine capacity fully.
Beyond storage, SiSense analytical power is very efficient at querying data. Our kernel runs in machines’ CPU and leverages the speed of L1 cache (about 50X faster than RAM). If you thought RAM was fast, think again!
Users get additional benefits from the fact that SiSense Prism is an end-to-end solution: it contains a database, ETL and visualizations in one box. This obviously yields deployment and cost benefits for customers, but it also results in better perform up and down the solution.
For the most technical of you, we’ve published more information about our technology here (prepare yourselves for some serious vectorization, super-scalar decompression and vertical fragmentation look-up!).
From a hardware standpoint, no big secret here. We can do this on ANY windows based machine. For the purpose of the Strata conference, we used a Dell machine – more specifically we used Dell PowerEdge R720 Rack Server with 160GB or RAM.
Price varies based on configuration and ordering options but the base model starts @ $8,200 and can be found here (by the way, you don’t have to use the same machine. We run on any Windows environment and are optimized for Intel and AMD chips).
Don’t believe it? Most people don’t until they actually try it. That’s why we make the full version of our software available as a free trial download here.