Will Snowflake spark a cloud data warehouse price war?

Snowflake, a cloud data warehouse pure play, has decided to slash prices on storage, and make its money on compute resources and management. Will this force Amazon, Microsoft and Google to respond in kind?
Written by Andrew Brust, Contributor

If the biggest impediment to the Data Warehouse market growth was cost and hardware scalability limits, then the logjam has broken. And if the biggest boon to Hadoop adoption was commodity compute and storage economics, along with elastic scaling, then things may be more complicated now.

Cloud-based data warehousing (DW), based on cloud storage and cloud-based clusters of database virtual machines, makes DW much easier than it was on-premises. And now, Snowflake Computing, a cloud DW startup, has slashed storage prices, possibly making the economics as attractive as the logistics.

Also read: Cloud data warehouse race heats up
Also read: Snowflake introduces multi-cluster data warehouse

Show me the money
here's the lowdown: Snowflake's terabyte/month storage costs are now $30 for "capacity" storage (sold on a contracted basis -- down from $125) and $50 for "on-demand" storage (sold on a pay-as-you-go basis -- down from $150).

Both price cuts are really big (76% and 67%, respectively). In fact, Snowflake says that it is essentially selling capacity storage at the same pricing customers would pay for Amazon's S3 (Simple Storage Service), on which Snowflake storage is based. And while the volume discount that Amazon gives to S3 customers is not being passed through, Snowflake says that is to keep pricing simple.

Provisioning gets easier too
In addition to on-demand storage, Snowflake is now offering an On Demand version of its entire service. With it, customers can now skip past formal sales engagements, hit the Snowflake site, open an account, provide credit card information, provision resources and be off to the races.

Of course, Microsoft (with its Azure SQL Data Warehouse service), Amazon Web Services (with its Redshift service) and Google (with its BigQuery service) already offer such self-service provisioning. But, according to Snowflake, those other services' storage prices are anywhere from twice to fifteen times as much.

Some of that math is based on Snowflake's storage compression ratios, which the company says are at least 3:1. With that compression rate, the $30/TB/Month cost of physical storage equates to $10/TB/month effective storage cost. (Compression ratios are content-dependent.)

Peer (pricing) pressure?
Will the three largest cloud providers need to respond in kind, with commensurate reductions in storage fees? On the one hand, the big guys could decide they don't need to respond to an upstart provider shouting to be heard in the market. But the cloud does tend to have an arms race-to-the-bottom habit when it comes to pricing. And as soon as one of three succumbs, it will be hard for the other two to abstain.

DW services like Snowflake's can be hugely attractive to customers who have been working with relational database technology for decades. But such downward pricing pressure in the cloud DW world could also have a spillover effect in the cloud Hadoop and Spark arena for big data.

Won't last forever, but...
Eventually the pricing will reach an equilibrium (i.e. a floor) and the providers will then monetize through premium features and services. But what's exciting right now is that the bar between basic and premium is moving north. And the providers -- none of whom wants to be in a low-margin business -- are having to innovate to keep revenue and profits healthy.

Either way that's a win for the customer.

Editorial standards