I would have benchmarked with an instance that has local nvme, like c8gd.4xlarge.

Do they make any promises about persistence of local NVMe after something like a full-region power outage yet? Because if you can't do durable commit on a single-region cluster that will be just temporarily unavailable without loosing committed data if something like that happened, it's not quite there unless you still stream a WAL to storage that they do promise you will survive a full blackout of all zones that store (part of) the data.

▲

amluto 3 hours ago | parent | next [-]

Yes. They promise to wipe your data. That SLA has all the nines you can ask for as long as you measure it in the right direction :)

▲

LunaSea 6 hours ago | parent | prev [-]

You already lose your data after instance restart so I think that full region outage is already out of question.

	▲	ody4242 4 hours ago \| parent [-]
		Idk how an AWS region would respond to a power outage, but i have tested this in AWS Outpost, and there, if you power down a rack, then power it back again, the baremetal instances will not be recreated. (I was surprised as I was expecting the EC2 health check to terminate them, but it does not work like that.) My understanding is that if you stop/start an instance, your local storage is gone (as the instance might even end up in a different host), but if you just reboot the instance, it should keep the local storage.

▲

devnotes77 5 hours ago | parent | prev [-]

Worth noting the c8gd local NVMe is ephemeral so you'd need to pre-stage the data each run, but for a benchmark like this that's actually ideal since you avoid EBS cold-read artifacts entirely.