21–30 of 35 posts
.. or go bare metal?
IMO:
- cloud is expensive with large resource needs, so bare metal is better
- cloud is hard to use for few deployment needs, so bare metal is better
So, the best case to use cloud is when you require lots of deployments but relatively small resource needs for each. For example, testing a piece of software on many platforms.
Cloud price comparisons are often apples to oranges. Due in part to an extremely large number of hidden factors. I'm not just talking about direct cost, but even 1vCPU != 1vCPU. When discussing s3/gcs storage; for instance, you would need to understand the difference between each of the storage classes and how they're charged for access. This would look much more like a series of a million graphs than a table. On the…
Generally for EC2 AWS is transparent about which CPU model and clock speed they are using. For other services, not so much. Like, is 1 CPU / Memory in Fargate the same as in App Runner and Lambda?
Hell, Brendan Gregg, father of observability, noticed a 10x performance difference of machines of the same type inside AWS and that’s before talking about different providers.
GCP is also quite transparent with their CPU capabilities. But I would really hesitate to say 1x2.2GHz vCPU in GCP is equal to 1x2.2GHz vCPU in AWS
I see no mention of bandwidth pricing. Some time ago, I evaluated the pricing for someone who was thinking of moving a web server from a traditional private server to AWS. It seemed that the monthly cost would be lower, until I looked at the bandwidth cost. The monthly price of that traditional private server, which included a fixed bandwidth limit, ended up being lower than what it would cost just for the bandwidth…
I see no mention of bandwidth pricing. Some time ago, I evaluated the pricing for someone who was thinking of moving a web server from a traditional private server to AWS. It seemed that the monthly cost would be lower, until I looked at the bandwidth cost. The monthly price of that traditional private server, which included a fixed bandwidth limit, ended up being lower than what it would cost just for the bandwidth…
One killer no one seems to notice is the bandwidth between availability zones. When you have a proper best practice cross AZ deployment it can be rather expensive. But still cheaper than switches and humans and cages and data centre contracts.
I see no mention of bandwidth pricing. Some time ago, I evaluated the pricing for someone who was thinking of moving a web server from a traditional private server to AWS. It seemed that the monthly cost would be lower, until I looked at the bandwidth cost. The monthly price of that traditional private server, which included a fixed bandwidth limit, ended up being lower than what it would cost just for the bandwidth…
I see no mention of bandwidth pricing. Some time ago, I evaluated the pricing for someone who was thinking of moving a web server from a traditional private server to AWS. It seemed that the monthly cost would be lower, until I looked at the bandwidth cost. The monthly price of that traditional private server, which included a fixed bandwidth limit, ended up being lower than what it would cost just for the bandwidth…
There's so many variables. I was looking into moving my stuff into the cloud to throw out some physical hard drives. I haven't committed to anything yet but I was surprised to see dropbox is somehow cheaper than AWS. Isn't dropbox just an entire extra company that makes a layer slapped on top of AWS?
https://www.computerworld.com/article/3044261/dropbox-quits-...
I see no mention of bandwidth pricing. Some time ago, I evaluated the pricing for someone who was thinking of moving a web server from a traditional private server to AWS. It seemed that the monthly cost would be lower, until I looked at the bandwidth cost. The monthly price of that traditional private server, which included a fixed bandwidth limit, ended up being lower than what it would cost just for the bandwidth…