The honest answer to your question about where the line is: it's when you start writing your own health checks, restart logic, and deployment orchestration on top of compose. That's the tipping point where you're maintaining infrastructure instead of just running apps.
We built Miren for exactly this middle ground. Push to the cluster, it handles image building, TLS, and deployment on any Linux server. No Dockerfile, no compose, no Kubernetes. Autoscaling is on by default. It's not that Kubernetes is wrong, it's that there should be something between compose scripts and a full cluster.
We're actively working on ways of deploying already ready services in docker files to the cluster, which would probably be what you're looking for initially unless the cluster you're doing a lot of development on the cluster.
If you're open, love to get your thoughts on https://miren.dev. We've doing similar things, but leaning into the small team aspects of these systems, along with giving folks an optional cloud tie in to help with auth, etc.
Totally, it's all about the primitives. I'm curious where exe.dev is gonna build on the the base, or just leave it up to folks to add all their own bespoke stuff to do containers, logs, etc.
The last 20 years has given us a lot of great primitives for folks to plug in, I think that lots of people don't want to wrangle those primitives, they just want to use them.
Hi, Evan Phoenix, Rubinius creator here. I just wanted to stop in and say thank you for this comment. It means a lot to me, even after all these years that Rubinius helped you out.
It's very true that things like timeouts are done better within the client library. But Templar can combine timeouts with caching, allowing an API to appear to still be active and returning something even when it's just very slow.
Because a cache is best when it's populated often, it make sense to put that into a service that many different processes on different machines can use.
Everything within templar can be done by adding capabilities to client libraries, no argument. But most client libraries don't provide these capabilities in one programming language, let alone many of them. By having the functional available as a service, Templar provides these capabilities fairly transparently (and even more transparently in the future when there is per/host and per/url configuration).
Templar does not do SSL termination because it does not terminate requests for your api. It's designed to mediate connections between you and any APIs that you use, protecting you against those APIs being unreliable.
Yup! That's on the todo list. Auto-caching, as I'm calling it, would figure out a how long to cache something for and then be able to later on use HEAD to check if the url in question has changed.
That cache time could be as long as 5 seconds, so that Templar is checking pretty often if the upstream has changed.
Templar is designed to be used within your own infrastructure on a trusted network, where you talk to it and it talks to the outside world.
An attacker would have to be already within your private network, and if they were, they could observe the traffic in plain text already.
I think the confusion is around where Templar sits in relation to your app making HTTP API calls and the services you want to talk to. I'm going to draw up a diagram to help explain this better.
Ah! Ok, I got it. Feels like the right way to handle this is the allow a blacklist to be defined via config file, then applied as request, something like "X-Templar-Blacklist: internal". The list would be a set of ip ranges and thus you'd have to construct the list so that the EC2 => EC2 problem doesn't crop up, but it's doable!
Looking at that CPAN module, it wouldn't be hard at all to add a blacklist to Templar via a config file that is checked. The EC2 => EC2 issue is probably the biggest stumbling block to getting what you want. I'll have to investigate that, it seems like that would only happen if internal to EC2 they do ICMP redirects...
A great question. Nginx is not typically configured as a normal proxy but it can certainly do it. A difference is that Templar gives you control of options applied to each request differently rather than the same the same to all.
From timeouts to caching, one Templar can be used for all different kinds of upstream APIs.
I agree, async responses are a great tool. It's a question of figuring out a good way of delivering them back to the client. Another option is a long poll endpoint within Templar that it can deliver a stream of responses on. AMQP is an easy option because it keeps Templar simple.
Fire-and-Forget is on the todo list. Since you're asking about it I'll bump it up. The later, Templar could have an option that tell it to only connect to public ips. Would that do it for ya?
Author here! That's certainly an interesting feature that could be added. I have on the future list to support sending responses back via AMQP. Your callback scheme isn't much different than that.
We built Miren for exactly this middle ground. Push to the cluster, it handles image building, TLS, and deployment on any Linux server. No Dockerfile, no compose, no Kubernetes. Autoscaling is on by default. It's not that Kubernetes is wrong, it's that there should be something between compose scripts and a full cluster.
We're actively working on ways of deploying already ready services in docker files to the cluster, which would probably be what you're looking for initially unless the cluster you're doing a lot of development on the cluster.