> "Read operations assume a data size of 4K or less; each additional 4K costs an additional operation. Write operations assume a data size of 1K or less. Notably, index writes count as entirely separate write operations; they are not included in the document’s 1K."
So many customers don't account for this and it up costing $$$ if your data model isn't a good fit. Cosmos even takes it further w. 1kb units (I have spent hours on Cosmos pricing and am still baffled on how to price a workload.) Although... it does incentivize decent data modeling practices which often lead to more performant apps.
I see this type of comment thrown around a bit, but a quick pass of the senior management suggests otherwise. https://newsroom.ibm.com/executive-bios? Not a whole lot of MBAs and a decent amount of engineers.
- Rometty (Chair): Technical degrees, held technical positions for first 10yrs at IBM, then in sales for the 90s, two decades working primarily with finance customers and she worked the PWC acquisition.
- Krishna (CEO): EE, Idea guy.
- Whitehurst (Pres): MBA
- Boville (SVP Cloud): Business degrees (UK)
- Browdy (SVP Legal): IP attorney
- Foster (SVP Services): Art degree, Accenture guy
You don't need MBA to have MBA mindset. A lot of computer science grads use their degree to get foot in the door and then move into management positions.
At IBM, to do anything you need to get permissions from 5 different semi-tech approvers. They have not done any real work in a while but read a few blog posts and come up with their own policies that contradicts each other. It is a pita to want to produce high quality code.
And that is just middle management. All those people higher up can define tech words but hardly anyone can actually explain the definition of those tech terms.
The loosening lens is a bit of strawman argument. I would hope someone at the level has experience with the business side of things... Pichai and Nadella weren't tapped to run their business straight after shipping a release.
Also, Howard joined in May (which is an important one...because this press release is about Cloud not about HR/patents.)
Whitehurst has a Computer Science degree (BS) and was using Linux on his personal computers in the 1990s and early 2000s, I think it's unfair to drop him in the MBA bucket.
Something about cherrypicking outliers compared to the droves of managers who haven't written code in 20 years and have been converted to the dogma of management
There is an amount of lock-in with any Cloud workload though right? IAM, billing, support? If you are bought into containers (which quite a lot of people are) it doesn't seem like incredible lock-in from technology PoV when Fargate and Cloud Run exist?
Seems more like Cloud Run to me. A modest amount of vendor lock-in if you're only interested in running compute workloads, and the usual ton of vendor lock-in that is inherent to any effective cloud usage if you want to go all in with IBM Cloud.
It actually impresses me so far. I've got a soft spot for IBM Cloud given that its origin is creating a public cloud out of open source projects and giving back to them. Their serverless offerings so far have been lackluster though. IBM Cloud Functions, based on Apache OpenWhisk, which they created and open-sourced, has a weird programming model.
But this seems like Cloud Run except it also has support for Docker images on any registry (not just GCR) and can run batch jobs up to 2 hours long. I'm going to keep an eye on it.
Appreciate the offer, but I wouldn't know what to ask yet. I'm still pretty junior in the grand scheme of things. I plan to just keep an eye on how it develops.
So many customers don't account for this and it up costing $$$ if your data model isn't a good fit. Cosmos even takes it further w. 1kb units (I have spent hours on Cosmos pricing and am still baffled on how to price a workload.) Although... it does incentivize decent data modeling practices which often lead to more performant apps.