There have definitely been some problems we’ve had with it. For us, it wasn’t just take it out of the box and it worked right away. We had to do a lot of configuration to get it working in the state we wanted. It was a lot of back and forth with Temporal.
First, we used the self-hosted version, so you can self-host it yourself. But that didn’t scale too well for us. But we migrated to the cloud at some point.
We mostly worked with it on the cloud, but it wasn’t all good right out of the box, even after we migrated to the cloud. We had certain issues, and some of those issues were because we didn’t know all the features it had to provide. Some of those issues, even Temporal, we didn’t know how to fix because we think they were due to the very large volume of work we were scheduling in a very short amount of time.
In our use case, because we do a lot of work with it every single day, we had to spend quite some time configuring it and getting it to work how we wanted. But that’s not necessarily the case for other user flows that use Temporal and have a very different traffic pattern. Those were much easier to get working well with minimal configuration.
I was not involved in the initial step. I was involved in subsequent steps, which did the transitioning between the self-hosted and the cloud setup, but the cloud setup was already there when I came. I took part in provisioning some new namespaces, but it was mostly ClickOps. From my perspective, getting started with Temporal Cloud was probably more work on the business and product side rather than on the engineering side. I think that on the engineering side, the amount of work you have to do to get your cloud account running is really minimal.
Three resources were involved in the migration process and the setup of the cloud environment.
From maintenance point of view, we have to maintain the workers. But from a maintenance point of view, it’s alright. It’s a SaaS solution, so Temporal do most of the hard work. One maintenance aspect is the certificate rotation. That’s really annoying to do.
The maintenance for the self-hosted cluster is much more complicated and one of the reasons we’ve migrated. But, that’s also due to the fact that, being a security company, we have to respect strict security guidelines, which means that we have to modify some of the images that they provided because they didn’t respect those guidelines yet. For example, Temporal images aren’t FIPS compliant, and we have to be FIPS compliant.