undefined | Better HN

0 pointssrrdev6mo ago0 comments

Oh man this is so true. In this sort of org, getting something fixed out-of-band takes a huge political effort (even a critical issue like having your client database exposed to the world).

0 comments

3 comments · 1 top-level

DrewADesign6mo ago· 2 in thread

While there were numerous problems with the big corporate structures I worked in decades ago where everything was done by silos of specialists, there were huge advantages. No matter where there was a security, performance, network, hardware, etc. issue, the internal support infrastructure had the specialist’s pagers and for a problem like this, the people fixing it would have been on a conference call until it was fixed. There was always a team of specialists to diagnose and test fixes, always available developers with the expertise to write fixes if necessary, always ops to monitor and execute things, always a person in charge to make sure it all got done, and everybody knew which department it was and how to reach them 24/7.

Now if you needed to develop something not-urgent that involved, say, the performance department, database department, and your own, hope you’ve got a few months to blow on conference calls and procedure documents.

For that industry it made sense though.

eru6mo ago

Interesting. Wouldn't the performance department have their fingers in all the pies anyway, too, or how was that handled?

DrewADesign6mo ago

Their job was specifically managing server resource allocation— as an IT role and not a dev role— in a completely standardized environment. Most applications were given a standard allotment of resources, and they only got involved if something was running out of ram, disk access was too slow, or something just seemed to be taking a lot longer than usual. If it seemed to be a network problem, or just a program crash, for example, they were never involved unless troubleshooting indicated it involved them. More often than not, I’d get a phone call telling me the system I was working on seemed to be heavy on the disk access or something, and they had already allotted it more to keep it stable, but I should check to make sure we weren’t doing something stupid.

Now that I think of it, I’ll bet a lot of companies have a system similar to this for their infrastructure… they just outsource it to AWS, Azure, Google, etc. and comparatively fly by the seat of their pants on the dev side. You could only scale that system down so much, I imagine.

j / k navigate · click thread line to collapse