Amazon unveils data-warehouse-as-a-service
- 29 November, 2012 06:11
Andy Jassy, head of Amazon.com’s cloud arm, Amazon Web Services, used his keynote at the inaugural AWS re:Invent developer conference in Las Vegas to unveil the latest addition to AWS’ cloud ecosystem: data-warehouse-as-a-service.
Jassy said that a data warehouse service had been a common request from AWS customers.
“We thought, what would a data warehouse look like if we did it the AWS way,” Jassy said. “Well, it would have to be really easy to get started, self-service, able to scale up massively. We wouldn’t have to want anyone to pay upfront if they didn’t want to; they could use it as they saw fit. It would have to be really fast and it had to be inexpensive.”
Jassy said the new Amazon Redshift service would work with enterprises’ existing business intelligence tools.
Businesses can access two types of Redshift nodes: A 2TB node with two virtual CPU cores and a 16TB node with 16 virtual CPU cores. Up to 100 Redshift nodes can be clustered, for datasets containing up to 1.6 petabytes of compressed data. On-demand pricing starts at US$0.85 per hour for a 2TB data warehouse. Enterprises can also reserve instances, dropping the price to $0.228 per hour.
Jassy said that AWS conducted testing with Amazon’s retail arm, which employed an “old guard data warehousing solution”. Amazon.com’s data warehouse had 32 nodes, 4.2TB of RAM, 1.6 petabytes of disk and cost the company “several million dollars a year”.
“They did a test and they basically took 2 billion rows of data and six of their most complex queries and they ran it on Redshift versus their old world data warehousing solution they’ve been using [for] the last several years,” Jassy said.
“And on two 16TB nodes on Redshift for $3.65 an hour — roughly the cost of a latte, which we drink a lot of in Seattle — and $32,000 a year they were able to achieve on all those queries at least 10 times faster performance. So instead of spending many millions of dollars, for Redshift they spend $32000 a year and end up with 10 times faster queries. It’s pretty game changing.”
Amazon Redshift is currently available as a limited preview, with a full release in “early 2013”, Jassy told the conference.
Rohan Pearce travelled to AWS re:Invent as a guest of Amazon Web Services.