5 ms·
Is it? In general at every place I've worked the vast majority of the work has come from launching & supporting new products. Existing infra does not really bre
by type-r 4y ago
Is it? In general at every place I've worked the vast majority of the work has come from launching & supporting new products. Existing infra does not really break that often, so if we just want Twitter as it is, that should be doable by 1/10th of the people. I'm not picking just on Twitter -- I feel this is the case broadly. Maintenance of a system that doesn't change is fairly straightforward as compared to innovation.
- afavour 4y ago> if we just want Twitter as it is Musk explicitly doesn't want that though, because he needs Twitter to earn a lot more money than it currently does.
- shaburn 4y agoFalse. He can pump profits keeping earnings flat by slashing costs. See probably removing 1/3 the cost with this single move.
- afavour 4y agoThe maths don't even begin to work out there. He just made Twitter take on $13bn of debt to finance his purchase, the company will owe around $1bn a year in interest payments that it didn't used to. The layoffs announced today probably won't even cover that extra interest.
- shaburn 4y agoThat's probably a billion in recurring expense of the worst kind. Debt can be satisfied with an equity issuance in the future post restructure.
- WoahNoun 4y agoThat doesn't account for the lost ad revenue due to the rise of controversial content.
- smt88 4y agoYou clearly don't have experience with infra serving millions of people in real time. Everything is hard and infra breaks all the time.
- shaburn 4y agoHow many people are actually involved in fixing Twitter's infrastucture? Must be in the hundreds. Maybe a thousand with their own data centers.
- type-r 4y agoIf your experience with services not under active development is that they are "breaking all the time," I'm sorry to say that you've worked only at places where the quality bar is very low.
- saagarjha 4y agoServices that aren't under active development break as the things above them change and they are put into environments they weren't in before. This is very normal and the reason you don't see problems happening because of this is either because you're moving slowly (which is a valid way to do things…but not typical) or your have an army of people keeping it all running smoothly.
- origin_path 4y agoAlternative take: I was at Google when they killed Reader. At the time they pulled the plug, Reader had already been destaffed for years and was kept going in one guy's 20% time. The infrastructure effectively ran itself - the underlying Google datacenter teams replaced machines when they died, everything else was automated. My experience there was pretty consistent with that: if you leave things alone, they won't break. Virtually all breakage was caused by changing things. On a well automated system repairing the physical components, rolling OS updates from time to time and so on, is not a huge amount of work.
- smt88 4y agoThe idea that Twitter is ever "not under active development" is ludicrous. Services with hundreds of millions of users, antagonistic spammers/hackers/countries, and advertisers spending millions are never in maintenance mode. You sound like you're trying to apply your understanding of lifestyle software businesses to one of the largest real-time communication networks in human history.
- andirk 4y agoI remember the original Ruby on Rails Twitter going down a lot. And their error page said something like "Reason: Too many whales".