3 ms·
I agree with this. When organizations move their stuff into the cloud, they justify it as driving down complexity and reducing cost -- we don't need to rent spa
by bane 4y ago
I agree with this. When organizations move their stuff into the cloud, they justify it as driving down complexity and reducing cost -- we don't need to rent space for a data center, hire people to build and maintain that stuff, and so on. But the costing models seem to assume that other burdens like billing and cost management will be handled by the current in-house team.
Except the billing for cloud services is byzantine and entirely unpredictable. So the org hires new specialists to manage this, and then new IT folks to setup and run some kind of CMP, and then they end up in a confused cost spiral that nobody can really explain but it certainly isn't saving anything.
The developers meanwhile really just appreciate all the prebuilt lego bits and relatively simple scaling which they could get in theory and on-prem, avoid the cost spiral, and get way better visibility into what costs what, where, and why.
The cost models get even more wrong when the org tries to go multicloud because its like nobody who builds those models understands egress costs.
- pid-1 4y ago> Except the billing for cloud services is byzantine and entirely unpredictable. I think AWS is the biggest offender in this area, by not enabling per resource usage by default.
- CharlesW 4y agoAren't all of these indictments of CTOs/COOs, rather than cloud providers?
- bane 4y agoYes. But I think in a sense, they're also somewhat handcuffed by the industry. Talent wants to work "in the cloud", and if you don't offer that you restrict your talent pool significantly. A big part of creating a desirable workplace is creating one in which people want to be hired into and in tech that really just means "the latest tech" regardless of it being a good idea or not. I think it's also part of the reason we're doing so little with so much in terms of efficient software. "The latest tech" also just means huge abstraction layers which eat up tons of compute but offer only easier development at the cost of much more expensive O&M. It's honestly part of the drive to the cloud as well as what we used to do on a few boxes with software written in C or a mainframe is now a scalable cloud architecture running mostly Python or some such.