Recently, the Harness team accepted an invite to the Spinnaker Summit in Seattle.
Spinnaker has great awareness among Continuous Delivery practitioners, and it was terrific to get a chance to meet the people behind Spinnaker as well as the teams actively using it.
What were our key takeaways?
1. It’s A Very Welcoming Community
Let’s be honest here: it was a little strange for Harness to be invited, as we are essentially an alternative to Spinnaker in terms of the fact that we compete for time and attention from the same teams. But we weren’t treated like second-class citizens. People welcomed us–even the Netflix and Google crew who spend a great deal of time promoting and evangelizing Spinnaker.
The general perception seemed to be that Continuous Delivery is an evolving space, and not all solutions are going to work for everyone, so multiple points-of-view were very welcome. We were anticipating a tough few days; in reality, it was a fun time spent with peers.
2. Organizations Still Struggling to set up Spinnaker
One thing that became very clear was that organizations are very interested in Spinnaker, but very few can plant a flag in the ground and declare “Victory.” Even some of the big name logos typically associated with Spinnaker admitted that they are only partially installed; many others reported that it took several months to get a basic working version in place.
People weren’t attending the conference to share success stories, generally, but rather to get tips and tricks on how to become operational. (This fact was evident that in the Q&A for the opening session, more than one question related to stability and usability.)
The Harness point of view on this is simple: set up needs to be fast. We typically set up in less than a day, and customers can start building complex pipelines quickly (and ultimately pipeline building will take minutes rather than days or weeks). Time-to-value is so, so important with the pressure that DevOps teams are under to automate and ship faster.
3. Spinnaker Requires Dedicated Resource & Teams
The teams that were successful with Spinnaker generally have the ability to train a power user and let him or her run a team. Even with those resources, it seemed to take 6 months or longer to get the platform up and running.
Organizations are dependent on these power users in order to ensure that the platform remains usable, and we also heard a lot of anxiety over what might happen once that power user decided to jump ship (the reality is that DevOps professionals will change companies several times over the course of their career).
If a CD platform is going to solve problems enterprise-wide, it can’t be a single person’s project; it needs to be democratized across not only the DevOps team, but even the line developers as well. If someone leaves, that person’s replacement needs to be slotted in quickly and easily.
At Harness, we’re trying to master Continuous Delivery as-a-Service where teams get 90% of the setup and capabilities out-of-the-box.
4. Other options for Continuous Delivery are actively being sought
People weren’t laser-focused on using Spinnaker. They were still considering all their options, including simply extending their Jenkins pipelines (not a lot of fun, but at least it’s the Devil You Know).
They also spoke to consultancies and professional service firms that promised oversight and implementation of Continuous Delivery strategies. And, yes, they talked a lot to Harness about a commercial solution that may have the advantages of a dedicated customer success team and a fast time-to-value.
5. Machine Learning and Security Still not part of the Conversation
Some of the tenets that Harness holds dear in regards to Continuous Delivery are really not part of the conversation. From our standpoint, Machine Learning is absolutely crucial to an enterprise-grade Continuous Delivery platform, including the ability to verify the success of production deployments by integrating with monitoring and logging tools, as well as the ability to automate production rollbacks.
Spinnaker does have Kayenta for automating Canary Services, but this only works with a Netflix tool called Atlas as well as Prometheus, Stackdriver, and Datadog. This is a far cry from how Harness uses neural nets and Kmeans clustering in order to automate all verification and rollback processes, not to mention the configuration of such.
In addition, financial service and healthcare companies need security baked into the CD process, which is why we have Role-Based Access control (RBAC), a fully documented audit trail, and secrets management–ensuring a CD platform that even the security team will respect.
Obviously, everyone approaches Continuous Delivery with their own perspective. There are organizations that will only use open-source tools, no matter what, full stop. And that’s fine–Spinnaker is there for them.
But for other organizations that are looking for as much simplicity and automation as possible in their enterprise-grade CD platform, Harness remains a great option. We love the fact that the Continuous Delivery space is growing and that there’s now more than one way to deploy. We truly believe that there are lots of great options to suit everyone (and doubtless more to come)–and the one you choose is less important than finding exactly the right way for your team to Get Ship Done!
If you’re interested in comparing Spinnaker vs. Harness, you can read our blog or you can sign up for your free trial and kick the tires.