It’s sort of amazing how fast the Google Cloud Platform has gone from not-really-there to a contender for Amazon Web Services’ public cloud mantle, at least in terms of perception.
This is not to say the gang in Seattle are shaking in their boots, but you can bet they keep an eye on what Google is doing. Going further, I’d wager that AWS is already considering sub-hour payment increments for its base level services, for which it now charges by the hour. (Google, Microsoft Windows Azure, ProfitBricks and CloudSigma already do this.)
The consensus from an unscientific survey of cloud watchers, is that Google is the one company with the engineering wherewithal and cloud focus to create and deliver innovative services on par with what AWS has done and continues to do.
Here are some things Google could do to shake things up for Amazon according to these folks (and myself.)
1: Launch reserved instances.
Other clouds can meet or beat AWS on price, at least when it comes to on-demand instances. But it’s hard for them to do so against Amazon EC2 reserved instances (RIs), which customers lock-in for periods of one or three years to get maximum discount. What would it cost Google to do the same? Or heck, even offer longer-term deals — say 5 or 10 years? (A little birdie tells me while Google does not list RIs on its price sheet, customers can get similar deals by contacting sales.)
2: Churn out more managed service-y services.
These are the sorts of things developers use but without having to twiddle all the underlying knobs. Google has already trotted out Google Cloud SQL (vs. Amazon RDS?) and Google Cloud Data Store (vs. Dynamo DB.) And don’t forget Google App Engine and it’s PaaS-like capabilities.
But it needs more such dev-friendly services if it wants to unseat AWS as the developers’ favorite cloud.
3: Parlay search.
Search is what Google is really, really good at. If it could manage to bring killer search to Cloud Storage, Cloud SQL and Cloud Datastore, “it will have one up on AWS,” said Gigaom analyst MSV Janakiram. Google Compute Engine will “eventually support shared storage with concurrent read/write ability which will be a big deal,” he added.
4: Get louder about cloud
No disrespect to Google Senior Vice President Urs Hölzle or director of product management Greg DeMichillie, but compared to AWS CTO Werner Vogels, ubiquitous blogger Jeff Barr, and AWS Senior Vice President Andrew Jassy, they’re just nowhere near as visible.
Google needs more — or at least louder — cloud evangelists. Oh, and more marketing.
5: Offer more and different types of instances
If Google could offer live instance resizing — admittedly a difficult technical task — it would be a huge differentiator, said Sebastian Stadl, CEO of Scalr, a cloud management company. He also thinks lowering the cost of on-demand instances to meet AWS reserved instances could be a game changer.
6: Add more regions
Google has more regional data center coverage at least in Europe and needs to publicize it. AWS, famously, has one European region operating out of Dublin, along with several edge locations while Google has data centers in Finland, Belgium and Ireland. I mean really, one AWS region for all of Europe? That sort of begs disbelief. That’s probably not going to be the status quo for long but in the meantime, Google has to toot its own horn.
7: Offer virtual machine image import/export
AWS already lets you pump your virtual machine images in and out of its cloud. If Google wants to win over some of those workloads, it needs to add that capability to its cloud. Easy entry (and easy exit) is a big thing in cloud.
8: Do all of the above but faster.
Google has rolled out a ton of stuff over the past year, but AWS has a seven-year head start, so Google Cloud needs to do more, and do it faster. The company has the technical chops and the resources to build all this stuff. Now it has to get cracking to deliver more of it ASAP.
One thing is true of Amazon Web Services: It’s a moving target.