SiSense announces analytics on Rackspace cloud

SiSense brings its single-server Prism BI/Big Data package to the Rackspace cloud, offering low-overhead analytics, on a subscription basis.

Disclosure: Readers should take careful note that SiSense is a client of my company, Blue Badge Insights. This article is not commissioned or compensated by SiSense, and I intend it as an objective report on the company's announcement.  However, my point of view is certainly subjective, and readers should bear that in mind.


Tel Aviv, Israel and Redwood Shores, CA-based SiSense announced today that it will offer its integrated analytics engine, data modeling and data visualization product, called Prism, as a hosted service on Rackspace's cloud platform. The announcment coincides with the opening of the GigaOM Structure conference in San Francisco, where the offering will be demonstrated today and tomorrow.

Prism's ElastiCube engine handles Terabyte-scale analytics and does so on a single server node, rather than a cluster of servers. The engine achieves this efficiency through a combination of technologies and techniques, which the company calls "in-chip analytics." The engine targets L1, L2 and L3 in-CPU cache, which have access times far smaller than that of a server's main memory, let alone those of solid state disk (SSD) and hard disk drive (HDD) storage.

SiSense's product combines the cache awareness with columnar compression, query recycling and the use of single instruction-multiple data (SIMD) CPU programming instructions, which process multiple data items together, rather than one at a time.

Since Prism uses these efficiencies to run on a single server, minimal cloud resources are required to run it. As such, the Rackspace-powered Prsim offering comes in at a low cost, relative to the amoumt of data it can process. In fact, SiSense claims that customers' net cost with the Rackspace-hosted solution is $1 per terabyte per hour.

Of course, that calculation assumes a highly utilized server over the course of a billable month, which may or may not reflect certain customers' usage patterns. But the figure is nonetheless competitive and significant. And while SiSense's non-clustered architecture means it can't take on the data volumes that other technologies, such as Apache Hadoop, can, the fact remains that many businesses' data volumes are well-matched to Prism's capabilities.

Those customers are likely to be the same ones that find the ease of cloud-based deployment so attractive, so the SiSense-Rackspace pairing makes sense, as both companies are looking to lower barriers to entry.  

Could analytics become as commonplace as, say, customer relationship management (CRM)? 2013 certainly seems to be a year in which great strides are being made in that direction. Now we just need to make writing MapReduce code as simple as entering a new sales opportunity.