Опять клевещут!!
Jun. 8th, 2018 08:48 pm![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
scif_yarI work for a large global consulting company as a director for the technical consultants in my region.
Over the past year I have seen a lot of DevOps disasters, this is not a rant for or against DevOps, mostly just some insights from our clients.
I've seen some companies where a DevOps methodology is a great fit (usually companies where their entire business is based on software development)
But I've seen some disasters as well.
We have quite a lot of clients who went full DevOps / Cloud (usually not from our consultation, generally went it on their own) and now it's become an unmitigated disaster, Requests to bring companies back from the cloud to on-premise data centers has increased six fold since the beginning of 2017, we quite literally don't have enough consultants to work on all these projects.
The biggest issue seems to be that most of these companies did not try to solve a problem they had with their IT department, but they just assumed it was a better way of operating, i.e. cheaper.
A lot of companies are learning now that they are not Google/Facebook and don't have the in-house engineering capability to do this properly.
Here is a general run down of the problems I've seen.
Most went from being IT departments of 10-20 people (70% helpdesk/desktop support, 20% sysadmins, 10% senior sysadmin type of structure) using proprietary enterprise systems that were managed in-house.
They transitioned into DevOps based methodology IT departments with usually 20% extra full time staff, mostly seniors being paid highly and the removal of many supported enterprise systems for equivalent open source projects.
The cost savings exercise of scrapping enterprise systems has failed in a lot of circumstances.
As an example a client we are working with right now had some software that was costing them $50,000 per year with 24/7 support, IT rarely needed to do much with this aside from occasionally troubleshooting issues.
Three years ago the VP pushed his technology department to scrap that software for an open source project, the problem was the open source project was completely bare bones and needed a lot of work customizing it to their requirements, he predicted 100 hours work.
So 3 years later and 6000+ hours of work from developers and DevOps staff, with almost all of the people involved earning over $150,000 per year and this "free" open source project has cost more than 7 years worth of support on the enterprise system.
It also only has about 30% of the features from the original system, so needs at least another 6000+ hours of development time.
Another major factor that is killing these forays into DevOps and cloud is the cost of cloud, it's almost always underestimated, sometimes grossly underestimated to the point of people who calculated the cost ended up losing their job for costing the companies millions in extra cloud usage.
https://www.reddit.com/r/sysadmin/comments/8phm0z/devops_horror_stories/
Over the past year I have seen a lot of DevOps disasters, this is not a rant for or against DevOps, mostly just some insights from our clients.
I've seen some companies where a DevOps methodology is a great fit (usually companies where their entire business is based on software development)
But I've seen some disasters as well.
We have quite a lot of clients who went full DevOps / Cloud (usually not from our consultation, generally went it on their own) and now it's become an unmitigated disaster, Requests to bring companies back from the cloud to on-premise data centers has increased six fold since the beginning of 2017, we quite literally don't have enough consultants to work on all these projects.
The biggest issue seems to be that most of these companies did not try to solve a problem they had with their IT department, but they just assumed it was a better way of operating, i.e. cheaper.
A lot of companies are learning now that they are not Google/Facebook and don't have the in-house engineering capability to do this properly.
Here is a general run down of the problems I've seen.
Most went from being IT departments of 10-20 people (70% helpdesk/desktop support, 20% sysadmins, 10% senior sysadmin type of structure) using proprietary enterprise systems that were managed in-house.
They transitioned into DevOps based methodology IT departments with usually 20% extra full time staff, mostly seniors being paid highly and the removal of many supported enterprise systems for equivalent open source projects.
The cost savings exercise of scrapping enterprise systems has failed in a lot of circumstances.
As an example a client we are working with right now had some software that was costing them $50,000 per year with 24/7 support, IT rarely needed to do much with this aside from occasionally troubleshooting issues.
Three years ago the VP pushed his technology department to scrap that software for an open source project, the problem was the open source project was completely bare bones and needed a lot of work customizing it to their requirements, he predicted 100 hours work.
So 3 years later and 6000+ hours of work from developers and DevOps staff, with almost all of the people involved earning over $150,000 per year and this "free" open source project has cost more than 7 years worth of support on the enterprise system.
It also only has about 30% of the features from the original system, so needs at least another 6000+ hours of development time.
Another major factor that is killing these forays into DevOps and cloud is the cost of cloud, it's almost always underestimated, sometimes grossly underestimated to the point of people who calculated the cost ended up losing their job for costing the companies millions in extra cloud usage.
https://www.reddit.com/r/sysadmin/comments/8phm0z/devops_horror_stories/
no subject
Date: 2018-06-09 05:32 am (UTC)- Мистер Макконнахи, это дело Йорков, которое тянулось годами, я его закрыл за неделю.
- Идиот! Пять лет дело кормило контору, пять лет! И еще столько же кормило бы, а он его... закрыл!
Ты еще учти такую вещь - чем дальше в лес, тем больше денег и времени спущено на облако-девопс-нужное подчеркнуть. Тем больше горит стул под принимавшим решение, и тем более отчаянно он за него цепляется: sunk costs fallacy, помноженное на осознание того, что человек вылетит из конторы впереди собственного визга. Постоянно переписываются цели, задачи и так далее, и может статься, что пару лет спустя начала проекта у него сменились все цели - ачетакова, "Agile".
Самое обидное - это когда спустя пару лет понимаешь, что если бы все эти человеко-часы были спущены тупо на улучшение уже имеющегося хозяйства, оно бы превратилось в такую конфетку, что никакого облака с девопсом не надо. Точнее как, если прищуриться, то имеющее хозяйство и так имело немало черт и облака, и девопса, и за эти сотни и тысячи человеко-часов эти черты можно было расширить и углубить до того, что отличия от облака и девопса были бы настолько косметическими, что на практике были бы совершенно незаметны.
no subject
Date: 2018-06-09 03:37 pm (UTC)no subject
Date: 2018-06-10 02:27 pm (UTC)no subject
Date: 2018-06-09 01:51 pm (UTC)no subject
Date: 2018-06-09 03:35 pm (UTC)