3 ms·
I have seen plenty of companies that hand roll their own ops/alerting system, and sure that is a fine strategy if you can put enough resources into it, but that
by paxys 4y ago
I have seen plenty of companies that hand roll their own ops/alerting system, and sure that is a fine strategy if you can put enough resources into it, but that is mostly never the case. Do you host it separately from the rest of your infra? Does it have multi-DC failovers? Does it have a dedicated SRE/on-call team? Do you load test it? Do you test that it can send texts to one of your developers who has bad coverage in the middle of India or Romania or Bali?
The problem usually is, when the rest of your systems are down then this one is likely to be down as well, making it useless. People are paying PagerDuty primarily for their uptime and reliability.
- ajoseps 4y agoi think these really depends on the requirements of the company. there will always be a long tail of scenarios to support, that doesn't mean every company needs support in all those scenarios. some things can be good enough
- paxys 4y agoAll of what I mentioned (except maybe international user support) is table stakes for an alerting system regardless of whether the company size is 1 or 100,000. A system that works 99% of the time is useless because it is only really needed for the other 1%.