Datallegro has grid-enabled its data warehousing appliance system and added a new online storage option that provides more flexible configurations for mixed-query environments.
Datallegro said that grid-enabling its v3 appliance gives customers the flexibility to implement a variety of hub-and-spoke and multi-temperature data warehousing architectures. That’s a change from traditionally rigid enterprise data warehouses that typically force companies to physically centralize their data in a single warehouse.
We’re the first appliance vendor to offer grid-enablement, claimed Stuart Frost, CEO of Datallegro. It’s natural for us to break up our architecture and it’s simple because of the way our appliances are made up of commodity server and storage components.
Frost said that Datallegro’s appliance, the guts of which include Dell servers, EMC storage, and Cisco switches, is fundamentally built to operate like a grid.
Our server and storage appliances are tied together by an InfiniBand [Server Switch] backplane and we use open source technologies to string together services like provisioning and message passing between them.
Customers can now build a hub-and-spoke grid of appliances that has an ability to transfer data quickly between linked nodes.
That wasn’t possible before in traditional federated architectures given the short overnight loading windows. Our appliances overcome that because they’re designed to move terabytes of data around in minutes.
Grid-enablement means that Datallegro can effectively partition off mixed processing requirements across a series of appliance-driven data marts, hubs, federation, and conformed dimensions. For example, a company can create appliance hubs for extract, transform, and load (ETL) or data governance.
The difference, Frost explained, between grid-enablement and stringing together appliances through standard ODBC interfaces is speed of data throughput and processing.
Our grid integrates multiple appliances through the same high-performance backplane, allowing data marts and hubs to be quickly populated and updated. We’re moving highly compressed data between our appliances in a massively parallel way.
Datallegro is offering offer grid-enabled versions for its multi-rack, single-rack, and encrypted data warehouse appliances.
Frost said the company only last week closed a huge deal at a large electronics manufacturer for a 270 terabyte capacity system comprising of 8 appliance nodes connected via a grid. According to Frost, that company has built a classic hub-and-spoke architecture with separate hubs for each business region, Asia Pacific, EMEA and North America, with spokes running off them to data marts. Those hubs are tied back to an overall corporate hub that has summaries of the data.
We expect to rack up large, multi-million dollar sales on the back of our new grid-capabilities, Frost said.
He also said the grid-architecture is highly suited for disaster recovery provisioning applications.
A not so apparent key benefit of the grid architecture is to support so-called multi-temperature architectures that blend hot appliance data that is more up to date and accessed frequently and cooler historical archived data that is accessed less so.
To help with these types of configurations, Datallegro now offers a new online storage offering is designed to store historical data that would normally be shoved onto a tape archive that is normally slow and expensive to access.
The online storage substantially speeds up access and lowers the cost by making up to 200 terabytes of data readily available for querying on a single rack.
We can do that for less than $8,000 per terabyte, Frost said.
The new online archive appliance is available in a standard option or as grid-enabled multiple racks linked together) to work within a multi-temperature or a standard hub-and-spoke architecture.
Importantly the archive is built on the same core Datallegro appliance components as well. Whereas other data warehousing vendors like Sand Technology address the problem by tacking on a fat archive system onto the back of the database, so that occasional queries can be directed to it without having to go to a slow tape library, Datallegro is different Frost claims.
We view the archive as an integral part of the data warehouse and it’s built on the same technology parts. What we’ve done is basically change the balance of Dell servers and EMC storage in our appliances in favor of storage.
We’ve also figured out under the hood where the queries will hit, making it seamless and transparent to users.
With regulatory and compliance issues now forcing companies to make a growing amount of their data accessible online, the online archiving capability provides companies with an easy and cheaper way of keeping it online without having to battle with the costs and performance issues of storing it in a mainstream data warehouse.
Datallegro has already sold the grid-enabled archive system to a large telecommunications provider, allowing it to store as much as 200 terabytes of call detail record data previously taped data online in an online archive for fraud and revenue assurance analysis. The system is expected to go live next month.
Frost said that Datallegro’s business has grown leaps and bounds over the past year. We’re 100 people right now and revenue topped 200% growth last year. This year we’re looking at even faster growth than that, and I wouldn’t be surprised if we hit 300%.
While Frost admitted that the company is not profitable right now, he said the company is on track to do so in the next 12 months.
Frost isn’t thinking of an IPO but said the company had been approached by financiers that missed the Netezza deal, a rival that IPO’d earlier this year.
In 12 to 18 months we might be in good shape [for an IPO]. But right now we’re heads-down focused on getting happy customers.
Frost said that close ties with influential vendors, notably EMC, has helped the company to extend its market reach.
Having a partner like EMC in our corner has helped us immensely, and we’ve already had a few wins with them. They’re taking use into big accounts at a very good level, as lead data warehousing partner. And we’re also helping them with their storage business as well.
It’s started really well but we still haven’t fully exploited the relationship.
Frost said to stay tuned for further announcements with EMC later this year.
Datallegro unveiled the new software products at the TDWI conference in San Diego, California today.
The debate over centralized versus federated data warehousing architectures has been given a new twist by the advent of grid-enabled data appliances.
Previously, performance and adaptability to business change used to be a daily thorn in the side of large centralized data warehouses. And the costs and data loading performance conststarints imposed on keeping data synched between dependent and independent data marts have plagued supposedly more nimble federated data warehouse architectures.
Datallegro’s flexible grid-enablement capabilities address both these issues by allowing a high-performance hub-and-spoke design that meets data loading speed, query performance, and stringent levels of data governance that companies expect today.
Given that most large data warehouses are hit by a mixed workload of constant and infrequent queries, with each dipping into current and archived data, it certainly makes architectural sense to store different types of data on different warehouse systems. The added benefit of doing this on an appliance is of course an attractive price-performance tradeoff.
Datallegro’s online storage capability is also testament to the benefits of using standard commodity server and storage components. It shows demonstrates how opting to use standard, as opposed to proprietary servers and storage, as most appliance vendors do these days, allows for greater product innovation.