First things first

In Sunday's newsletter, I asked whether IaaS can continue to be a zero-sum game, where customers pick a cloud provider and go all-in with it. The question was rooted in the contentious Pentagon cloud contract currently up for bidding (more on that in a bit), but it definitely applied more broadly. And in the past few days, we've seen more evidence emerge that a multi-cloud future really might be in the cards.

Cases in point:

On the storage front, an astute reader also reminded me that Apple has been using both Google and AWS for iCloud storage, and it's my understanding that Backblaze is also winning customers for its B2 service that previously stored solely with a single cloud provider. (Backblaze also partnered with some IaaS providers recently to try and decentralize IaaS compute and storage.)

But the Netflix-Google relationship is to me the bellwether here, despite Netflix's claims that there's nothing to see other than some disaster recovery workloads. That may be true, but it's also true that Netflix and Google just partnered on a code-analysis tool called Kayenta, which is based on the open source Spinnaker CI/CD platform created at Netflix and since embraced by Google. It's at least a little interesting that Google is working with Netflix to help advance some of its open source projects (Spinnaker is now even building something of an ecosystem around itself) when it never appeared, at least publicly, that AWS was too involved. (There's also that whole Netflix vs. Amazon Video angle ...)

Last year, I argued that Google could distinguish itself as the open cloud platform and establish a reputation that could go a long way in winning customers that truly care about using open source technologies even in the cloud. Neither Kayenta nor Spinnaker will likely ever reach Kubernetes or TensorFlow levels of popularity, but a large stable of quality open source cloud projects could certainly help Google compete against AWS. Even if it would prefer its customers not being doing multi-cloud, a win is a win, and the perception that Google supports multi-cloud workloads could help it score more of them.

One more thing: the Netflix-Google piece linked to above notes that Netflix actually has been using Google for artificial intelligence workloads, on top of just DR storage. This wouldn't surprise me a bit if it's true. For one, Google is the only cloud provider presently offering customized AI hardware at the moment, with its Cloud TPUs that are specialized for TensorFlow jobs. Also, I recently interviewed an exec at a publicly traded company for the ARCHITECHT Show podcast (stay tuned to find out who!), who pointed out that although his company runs on AWS, it's always evaluating its options and it does like what Google is doing in AI.

However ... Netflix also did open source its Titus container management platform this week (one of its biggest open source releases in quite a while) and commented in the blog post how its tight AWS integration helps Titus perform so well. Still, it's possible Netflix is only releasing Titus now because it's already moving onto something newer (and perhaps less AWS-centric) or that any workloads for which it's using Google are distinct from those that Titus manages. Or, I guess it's also possible that Netflix is still committed to AWS for the foreseeable future and there's just an awful lot of smoke ...

For more on Google's cloud momentum, largely around AI and even multi-cloud deployments, check out this newsletter from back in August.

Now, a few more things.

More drama in the Pentagon cloud deal

Microsoft appears to have a strong future in IoT

Maybe IBM's future is not in cloud computing, per se


The ARCHITECHT Show podcast

AI and machine learning

Sponsor: MongoDB

Cloud and infrastructure

Sponsor: Replicated

Data and analytics