How Google and Amazon think about multi-cloud workloads

It was not a big announcement from Google this week, but in a way it could be seen as a symbolic one.

In a blog post, Google announced new functionality that will allow its Cloud Endpoints API (Application Programming Interface) management platform to integrate with Amazon Web Service’s Lambda functions-as-a-service product. The idea is you can build an application in Lambda and use Google Cloud’s Endpoints to manage API calls associated with it.

+MORE AT NETWORK WORLD: Serverless explainer: The next generation of cloud infrastructure +

It’s a refreshing reminder that Google is willing and open to talking about customers using other vendor’s cloud products and services, including from their biggest rival, AWS.

Google certainly isn’t cozying up to AWS. It’s cloud chief Diane Greene this week put AWS in her crosshairs, predicting Google will top AWS within five years.

When asked for comment, a Google spokesperson pointed out other ways that Google helps customers manage multiple clouds. It has a product named StackDriver that allows customers to monitor and log their use of not just Google’s cloud but AWS too, for example.

AWS has somewhat of a different story when it comes to hybrid and multi-clouds. For the first few years of the company’s re:Invent conference, it spoke about how in the fullness of time most workloads would end up in the public cloud. In recent years, however, it has introduced tools that allow customers to connect their on-premises data centers or co-location facilities to the AWS public cloud. Products like Direct Connect, Amazon Virtual Private Cloud and the company’s AWS Storage Gateway all are examples. It also recently announced a partnership with VMware that will allow workloads running in that software to run in AWS’s cloud more easily.

At the recent AWS Summit in San Francisco CEO Andy Jassy said he believes hybrid clouds play an important role for customers (he starts talking about it around the 8:30 minute mark in this video), but he says when push comes to shove, most enterprises and CIOs tend to choose to standardize on one cloud platform. There are multiple reasons for this, he says: Splitting workloads across clouds can force companies to standardize on the “lowest common denominator” across cloud providers. Jassy argues that AWS is in a “radically different” stage of maturity in its products and services and is iterating more quickly than competitors. “Most folks don’t want to tie the hands of developers behind their backs in the name of splitting across clouds,” Jassy says. Learning multiple platforms is a “pain in the butt” too, Jassy says. Trying to make the shift from on-premises to the cloud is hard enough, and learning multiple cloud platforms is an added, unnecessary layer of complexity, he says. Thirdly, he says customers diminish their buying power when spreading workloads across clouds because vendors offer volume-buying discounts, which are reduced if you don’t run as many workloads in their clouds. If customers want to use multiple clouds for backup or redundancy, that can be done by using a multi-region approach all in AWS’s cloud. “When most CIOs and enterprises look at this carefully, they don’t actually end up splitting (workloads) relatively evenly,” Jassy said. “They predominantly pick one (vendor); others pick one and do a little bit with a second.”

The perception is that Google is more open to multi-cloud workloads though, says Mike Kavis, an enterprise architect with consultant Cloud Technology Partners.

“Google’s the most open to being multi-cloud, it’s just in their DNA,” he says. “Almost everything they do is open source.” Most of the Amazon and Microsoft “hybrid” tools, Kavis says, are basically on-ramps to get more data and workloads into their clouds. AWS would point out that their hybrid cloud tools allow for ingress into AWS’s cloud to egress out of it, easing customers’ lock-in concerns.

To an extent, this should not be surprising. Google has a heritage of being open source and given its market position of trailing AWS in market share – which Greene readily admits- it makes sense that Google would want to make it easy for customers to transfer workloads from AWS into Google. AWS, for its part, is growing healthily and has not yet seen a reason to build integrations with other public clouds. In the meantime, if customers really want to pursue a multi-cloud strategy, perhaps their best bet is relying on a third-party integration tool and not native tools from cloud providers.