Google puts Amazon on notice with new Cloud Platform features

Google Cloud Platform now includes revamped pricing, better testing and deployment tools, and expanded VM support -- all to displace Amazon

Page 2 of 2

A cheaper cloud for all
Throughout the announcement, it wasn't hard to detect veiled jabs at Amazon's cloud services. Complexity of pricing was the biggest and most obvious one, with the presenters (among them Urs Hölzle, Senior Vice President of Google) noting how confusion over prepayments, or using on-demand instances versus flexible ones, have made cloud pricing more of a headache than it should be.

Google's idea is that  cloud computing should track the cost of physical computing more closely. Moore's Law has caused the cost of hardware to drop year over year, but cloud computing's dip in pricing hasn't tracked the hardware cost curve as closely. To that end, Google has cut costs 30 to 85 percent across the board for Compute Engine, App Engine, Cloud Storage, and BigQuery. The cost of a single-core U.S.-hosted instance type has dropped from 10.4 cents to 7.0 cents per hour, with the high end (16 cores) down from $1.659 to $1.120.

Google introduced Sustained-Use Discounts as a cost-cutting measure. Any VM used for more than a certain percentage of the month is billed at an incrementally discounted rate, with the price cut ranging from 40 to 80 percent of the base rate -- a net monthly discount of 20 percent, according to Google's math. "No upfront payments, no lock-in, and no need to predict future use," says Google in its announcement post -- easily read as direct pokes at Amazon's maze of purchasing options, including spot and reserved instances. The next move is clearly Amazon's, as it has cut prices many times in response to competition.

Big data? Yes, and fast data, too
Back when Amazon introduced Kinesis, its platform for real-time streamed data acquisition, it was clearly Amazon's way of horning in on the market for big data by way of the burgeoning Internet of things.

Now Google has taken a step of its own in that direction, with an addition to its BigQuery data analytics platform, named BigQuery Streaming. It's designed to ingest up to 100,000 records per second per table, with a cost of $20,000 per month for reserved queries and $5 per terabyte for 5GBps on-demand queries.

Google's demonstration of BigQuery Streaming involved running live queries on a simulation of incoming live data -- 400,000 power meters, each generating one event per minute, to simulate the city of Seattle. Among the BigQuery Streaming features touted by Google was that it eliminated the need for ingestion processing, warehousing, or sharding of the data; one could simply pipe it right in and begin analyzing.

The big competition here is not just from Amazon, but also from dedicated hardware vendors who insist that big iron is what's needed for big data, as well as software makers looking to turn batch-oriented, distributed data processing solutions like Hadoop into streaming data processing systems.

Google's been doing its best to put Amazon on notice for some time, not merely by offering a cheaper product, but also a more flexible and genuinely useful one. App Engine was the main vehicle for those changes, at least as far as developers were concerned. But now Google's interested in making the whole of its cloud offering -- not select pieces of it that appeal to a select audience (say, only PHP developers, or only those running hosted apps) -- into a force to be reckoned with.

This story, "Google puts Amazon on notice with new Cloud Platform features," was originally published at Get the first word on what the important tech news really means with the InfoWorld Tech Watch blog. For the latest developments in business technology news, follow on Twitter.

| 1 2 Page 2