4 ms·
Unfortunately, EC2 is not an option if we want to deliver 1 minute promise. I need VMs that I can flash start with all the software needed to run the tests imme
by punkpeye 4y ago
Unfortunately, EC2 is not an option if we want to deliver 1 minute promise. I need VMs that I can flash start with all the software needed to run the tests immediately available. The best I was able to achieve is 20 seconds using my current provider. However, they charge in hour increments.
That being said, I am constantly experimenting how to reduce costs. I think the final result is going to be a fairly complicated orchestration of pre-booted machines that are rented by month and those that we spin up ad-hoc to handle demand fluctuation.
- banana_giraffe 4y agoWould Lambda be an option? Not sure what the stack looks like here, but a quick Google suggests people have gotten Cypress to run under Lambda, at least.
- janosdebugs 4y agoIn my experience Lambda can have a pretty slow start when it's cold.
- banana_giraffe 4y agoSure, for an interactive endpoint, but for this a cold start of 200-500ms, or even a second or two for a Docker backed Lambda, sounds very workable.
- janosdebugs 4y agoYou could always be pre-pooling VMs once you hit a certain scale.
- zmgsabst 4y agoEC2 bare metal + running your own VMs would solve the latency issue; but now you’re going to have more complex rental dynamics. That would be $0.05 per core for an hour… or $0.05 for 60 copies for a minute. (Price for m5.metal on demand; a1.metal is half the price; m6a.metal is around $0.04) I think the idea is cool — and wish you the best of luck. I’m just trying to share why as a senior SDE your product doesn’t make sense to me: 100 integration tests a day is a fairly substantial team… and you’re charging a lot for it, at $2-3k per month.
- yuppie_scum 4y agoY’all need Kubernetes.
- CoolCold 4y agoSounds like you are on Hetzner. If that's so, you can take sever for around 60$ month there and put some load there - I doubt it gonna dip your pocket for such test.