Netflix Releases Spinnaker, Goes Multi-Cloud With Google, Microsoft
Updated 9am 18 Nov 2015, UTC+10. See below.
Netflix, the streaming video company, has released a tool called Spinnaker that it says “is an open source multi-cloud Continuous Delivery platform for releasing software changes with high velocity and confidence.”
EMC’s Pivotal to build Spinnaker over the past year or so, and have released the platform code on GitHub.
Spinnaker supports deployment to AWS, Microsoft Azure “actively underway”. There are also plans to add support for containers “in coming months”.
This is an interesting announcement for several reasons.
Netflix is also mentioned when AWS has an outage, because they’re highly visible when affected.
But if Netflix puts all of their eggs in the AWS basket, they are creating risk for themselves. Single-sourcing important parts of your supply chain places your revenue at risk, which can sometimes be worth it. Some companies mitigate this risk by forming close partnerships with their key suppliers, almost becoming vertically integrated, so that they can better influence those suppliers’ risk management. Others prefer a portfolio approach, where risk is reduced through multi-sourcing. If one suppliers has issues, you have others to choose from.
That appears to be what Netflix.
Netflix can now treat its cloud services more like a commodity, then they can simply move their workloads to someone else. This is quite a challenge in the cloud world today, hence the oft-used “Hotel California” analogy.
Given that difficulty, having Spinnaker would provide a competitive advantage to Netflix. If it can do this and its competitors can’t, Netflix can enjoy a cost advantage (through price competition) and a revenue advantage (fewer outages->higher perceived quality). So why make Spinnaker available to others? Wouldn’t you want to keep it to yourself?
Netflix benefits by having a larger, more robust ecosystem of standardised services it depends on. It can rely on competitive forces to improve the services and prices in key parts of its supply chain. It is in Netflix’s own best interests to ensure that AWS doesn’t become a de-facto monopoly who can then dictate terms.
This is where the partnerships with Microsoft, and Pivotal come in. These companies are keen to have good chunks of the market, so by making it easier for potential customers to use their services as well as AWS, they improve their position by weakening that of AWS. If they believe they are better able to compete in other areas (perhaps a perception of enterprise capability, or integration with other systems) then it makes sense for them to commodify this part of their offerings in order to commodify AWS.
The price reductions achievable through AWS competing with Azure, Google, and Pivotal are likely better than what Netflix could achieve through tight partnership with AWS and bulk purchasing. If Netflix cares more about reliable supply and price than other ‘value-added’ services, then this again makes sense from their perspective.
By having Netflix be the poster-child for this cross-cloud, multi-cloud ability, the coalition of companies hopes that others will be interested and copy them, thus boosting adoption of the Spinnaker approach at the very least.
Now we get to see if this tactic will pay off.
A spokesperson for Netflix emailed me to say that Netflix isn’t looking at other cloud services, apparently. Here is the email quoted verbatim:
> Hi Justin, > > Nice to e-meet you. Saw your piece and just wanted to give you a bit more detail since it’s actually untrue that we’re looking at using other cloud services. > > The implementation of this tool is to innovate and better serve our AWS needs, providing continuous delivery and faster deployments. Netflix leverages AWS for our cloud services and we have no plans to move away from AWS or to use Spinnaker with other cloud providers. > > The advantage of open sourcing this tool is that everyone can benefit from others’ innovations, even those that do not operate on AWS. We had a learning from our previous use of Asgard that many companies used this tool and ended up forking the code to use for their own cloud purposes and we lost the innovation when they forked. So creating a new tool that was usable across multiple providers will alleviate that problem in the future. > > Would greatly appreciate if you could clarify.
I’m attempting to talk to Netflix about this in a bit more detail, but timezones make life a challenge sometimes from down here in Australia.
It seems that Netflix would prefer that people continue to build on this tool out in the open so it can get the benefit of any good ideas other people have. Of course, if other people want to fork the code and use it privately, there’s still nothing to stop them. A private fork always has issues with merging in changes from upstream, but this isn’t a new issue compared to Asgard.
Perhaps Netflix really is an altruistic company who just want to make the world a better place, and isn’t playing the smart strategic game I gave it credit for.
Time will tell.
This article first appeared in Forbes.com here.