2 ms·
What I'm curious about is why it is a single-PaaS; I'd have expected Salesforce to have the customers quite isolated so the chance of bringing down multiple cus
by trebligdivad 18d ago
What I'm curious about is why it is a single-PaaS; I'd have expected Salesforce to have the customers quite isolated so the chance of bringing down multiple customers at once was much smaller.
- prpl 18d agoa few things are global (login service) to some extent, just as AWS places several such things in us-east-1
- kakwa_ 18d agoYou still have fleet wide management which can cause issues. Plus there are always a few core services like queues, authz, presentation layer. Also, mono-tenant architectures is no golden bullet either. Such architecture (often coming from a formerly on-premise product that was SaaS-ified) can easily become hell to operate as it multiplies the integration points (DB parameters, URLs, allowlists, etc). It's also quite wasteful in terms of resource utilization and hosting costs.
- stmw 18d agoThe customers are quite isolated, but it doesn't mean that some services or errors do not propagate. In public cloud terms, think back on some AWS or Azure or even Gmail outages - you probably wouldn't even hear about them if it didn't affect millions of users at once, across security and availability boundaries.
- Cthulhu_ 17d agoI think it's easy to underestimate how many smaller outages there are in any period of time but which do not affect anyone or only a small number of people due to all the mitigations, due diligence that developers and SREs do, and the self-repairing nature of modern systems.