A new generation of start-up enterprises has arrived in the business world. Riding a crest of maturing technologies, they are deploying new business funding and models in their mission to turn big-data into big-money.
The primary drivers have been the advent of near universal high-speed connectivity, in-memory databases, and ultra-low-cost memory and processing. As well as bringing the cloud into play, these have accelerated the penetration of social networking models into the business world, speeding up interaction and flattening organizations. The target markets for new model enterprises are industries needing ultra-high-speed, but reliable, transaction and analytic processing on gargantuan data volumes, such as finance, social networking, telecommunications, ad-serving, or online gaming.
For instance, there's GoodData Corporation of San Francisco, which describes itself as the LinkedIn of business intelligence. With only 180 staff on three continents, laptops all round, and no infrastructure, GoodData provides 6,000 customers with real-time business intelligence and analytics, letting them know what happened in the last hour, not just the last quarter.
Its services come as a platform in which the customer's data is warehoused on Amazon Web Services (AWS), and organized with in-memory database software. It is then mined, interrogated, and the results presented with GoodData's own software. Founder and CEO Roman Stanek estimates that without these new technologies, it would take five staff per customer, i.e. 30,00 staff, to do what 180 do today.
VoltDB of Billerica, MA claims to be a leader in the in-memory field, and describes its product as the world's fastest OLTP system. Aimed at a new generation of database applications -- real-time feeds, sensor-driven data streams, micro-transactions, etc. -- VoltDB scales linearly and the company reports performance of up to 3.4 million transactions per second on 30 nodes.
VoltDB is the brainchild of database guru Michael Stonebraker. He is also CTO of Paradigm4, another database start-up whose product SciDB features a massively parallel non-relational database integrated into an analytics platform capable of operating on datasets of billions of elements. This enables it to perform complex analytics in real time for industries and researchers with high-volume data inputs such as bioinformatics, finance, health care, insurance, or pharmaceuticals.
In advertising, targeted marketing is becoming more and more precise as larger volumes of customer data become available. But getting the value out of that data, and getting to the customer while it's still fresh, demands a combination of fast processing power and speed. Rocket Fuel Inc. of Redwood City, Calif., is a start-up that appears to be providing the goods.
Its business is "programmatic buying" -- real-time, precisely calculated targeting of ads at the most responsive prospects -- through some 26 billion impressions a day. Although shy about revealing its technology or results, it is, according to Forbes, genuinely growing ahead of the market in terms of customer acquisition, and has doubled its staff in the last year.
With new kids on the block displaying bleeding-edge technologies, with results to match, what are the big boys doing?
Microsoft has announced Project Hekaton (from the Greek for "100" and indicating expected increases in speed), which will bring in-memory OLTP capabilities to the next major release of SQL Server. This may not be until 2014, however.
ERP giant SAP is further ahead, having released its in-memory database HANA in 2011. It went further ahead last November when it announced a tie-up with VMware with vSphere as the preferred way to virtualize HANA's high-performance. SAP appears to believe that HANA's high-performance and vastly reduced development times put it well ahead of the market.
All this seems to leave at least one giant possibly exposed to cold winds of innovation. Oracle, while pushing the in-memory capabilities of its TimesTen and Exalytics products, does not appear to be wholeheartedly embracing the new technologies, and for the moment is clinging to its hardware heavy (it has to shift all those Sun machines), bolt-on software policies.
Could a few butterflies fanning their wings give a giant cold?