While this absolutely rings true today this is slowly becoming a non-issue as seeverless platforms migrate from FaaS to “run a container in a provider managed cluster.”
Also, you don’t have to “go all in on cloud” even with serverless if you care about portability. Bite the performance bullet, attach elastic IPs to your Lambda’s, expose them to your VPC and manage your own logs, metrics, auth, and networking. Trivial migration to EC2 if they ever stop serving you. And use service gateway for your managed services for another easy migration path if you ever want to switch to self-managed.
Half this article is absolutely right about the crazy amount of provider specific crap is in every cloud but the other half is presenting the normal operational burden of connecting disparate systems as something that is the unique fault of the cloud.
Also, you don’t have to “go all in on cloud” even with serverless if you care about portability. Bite the performance bullet, attach elastic IPs to your Lambda’s, expose them to your VPC and manage your own logs, metrics, auth, and networking. Trivial migration to EC2 if they ever stop serving you. And use service gateway for your managed services for another easy migration path if you ever want to switch to self-managed.
Half this article is absolutely right about the crazy amount of provider specific crap is in every cloud but the other half is presenting the normal operational burden of connecting disparate systems as something that is the unique fault of the cloud.