Open Source Journal

Infobright, the Toronto-based open source data warehousing company, is pushing out rev 3.1 of its low-cost software. It adds x86 Solaris 10 support to its Enterprise Edition and an enhanced SQL Framework to both its Enterprise and Community Editions so the widgetry should appeal to a wider swath of enterprises.

The new SQL Framework delivers better than 10x query performance on the complex queries that rely on SQL functions by way of the 100 additional SQL functions that Infobright has added to its patent-pending Knowledge Grid architecture.

The rev also upgrades the underlying MySQL database to release 5.1.

The company hiked up its skirt high enough to show off the Simplified Scalability architecture it has coming in Q4 as an alternative to complex MPP-based data warehouses. The widgetry, which currently works on a single server, will scale to 100TB and 300 concurrent users on 10 servers, according to marketing VP Susan Davis.

Only about 6% or 7% of the data warehouses out there are bigger than 20TB so where it's going is pretty ethereal. The upcoming capacity should reassure customers in light of the data explosion.

It promises pluggable extensibility so servers can be easily added or removed from the configuration to right-size for the load; near linear scalability; and automatic failover.

It leverages the company's data compression and a P2P server architecture. It follows a shared everything scheme in which every server will have access to all the data and there's no single point of failure.

Unlike pricey MPP architectures, data isn't duplicated or transported between nodes, which can cause bottlenecks. Because there's just a single instance of the data, storage requirements are reduced and implementation and configuration simplified. The solution includes a proxy/load balancer and management services.

Infobright claims its software-only solution offers the industry's lowest TCO along with operational simplicity. Its Knowledge Grid, its secret sauce, is supposed to reduce implementation and management efforts by up to 90%. It uses information about the data to minimize the need to decompress the data in response to a query. The company says other warehouses use indexing, tuning, and hardware for performance.

Infobright warehouses can be deployed on any x86 servers and various storage solutions including cloud deployment.

At the beginning of 2008 the start-up had maybe a half-dozen enterprise customers; now it's got about 50, Davis said. And since the company went open source last September, there have been 10,000 downloads. It claims to be the only enterprise-grade open source data warehouse. A subscription runs $10,000 per TB a year including support. It says it's the lowest-cost technology in the market.

The current rev gets 32 concurrent users and up to 50TB on a single server with 10:1 or better compression. It can load 300GB an hour and handles fast queries without partitions or indexing. High-availability storage is available with Sun's OpenStorage. There's Windows support in the Community Edition.

The stuff will work with any BI program.