|
Egress traffic cost is massively higher on both AWS and GCP. Instances of comparable power are somehow more expensive on both AWS and GCP. Also, simplicity; AWS IAM is mightily complicated, things like Cloud Formation are totally non-trivial, etc. You can get going more easily with simple and moderately complex setups on Linode or DO. Of course, AWS, GCP, and Azure have much bigger infrastructure, several availability zones, a lot of managed software (object storage, various databases, queues, email gateways, docker hubs, etc) which smaller players don't provide, or can't provide at the fault tolerance level which big players are able to offer. Something like AWS Aurora is hugely internally redundant to withstand link problems, node outages, etc transparently. If you want a thing like that, managing it yourself takes serious chops, and money. |