Re: out
Re DR, I would not be so black and white, each workload has an RPO and RTO and some workloads will be fine multi-az and others need multi region. Do you really require your admin server to be replicated to another region, or can you save the backups to s3 and replicate the backups to another region. How about your SQL server, maybe you need to replicate that to another region, but maybe your DR node can be sized small with high performance disks that can be scaled up on demand. Lots of options and I would say that one policy for DR would be too simple.
Re costs, it’s rare for an Admin/ITpro/anyone other than finance to have the end to end visibility of what it costs to run an on prem data Center. And until you close your data centre entirely, you end up running a DC and paying AWS to run theirs. And if your running 1000’s of VMs that’s not a good choice in my opinion. A lot of companies use AWS for serverless architecture and keep VMs on prem, seems to be a good strategy. I worked with a company who’s monthly asset tracking system built on serverless was never going to get out of the free tier, and the cio couldn’t believe he wasn’t going to have to pay anything. I worked with another company that went 100% all in, ditched all their onprem licences to go pay as you go and ended up saving about 50% of their total IT budget in year 1. So I guess it’s difficult to say oneway or another what’s best, but just comparing support costs and then writing of the cloud isn’t a fare comparison in my opinion.
Biting the hand that feeds IT
