Hacker News Reader

Comment by gnicholas | Hacker News Reader

The communications around this outage have been a complete mess. PG&E claimed the outage was being driven by high temperatures, low humidity, and high winds. My town, near Redwood City, was supposed to lose power today at noon. This was perplexing to us because it was not hot (75 degrees), dry (40+ percent humidity), or windy (6 mph).

But everyone has been running around preparing for the outage that we were assured would come. They’ve now pushed back the outage to midnight, but I have no idea whether that will actually happen.

And of course, their website has been down pretty much the entire day, and the phone lines are clogged. At the very least, they should have made sure the website could handle the load. They’ve been talking about the possibility of this shutdown for months, so there’s no excuse for the website being completely unprepared.

View on HN

The fact that all these corporations with billion-dollar budgets cannot make their websites work I'll never understand.

California population is 40 million folks.

It's trivial to have a single dedicated webserver handle 10k requests per second as a starting point.

That means that it'll take 4000 seconds to handle everyone's requests in aggregate, should they be made at the very same time — that's just a little over an hour.

Of course, folks won't make requests at exactly same time, and we aren't limited to a single server, either, and each server is actually capable of handling way more than 10k req/sec in the first place, after some optimisation.

E.g., for 50 bucks a month, anyone can easily run a service where, on average, every single resident of California can get info from a webpage about the current status once every hour without any delay. Simple round-robin DNS, and for 500 bucks per month, we can handle 10 requests per hour from every single resident. Etc. (Keep in mind that a brief search indicates a single server is actually capable of handling 500k req/sec, not just 10k, so, I'm being very conservative here with the 10k number.)

Yet all these companies spends millions of dollars on their IT infrastructure and teams, procure all sorts of external CDNs, and yet cannot make their website perform to these most basic specs.

Why? Alas, that's because they gotta have all those fancy graphics, the tracking for user experience, and all the dynamically-generated and uncacheable webpages that costs millions of dollars to maintain and which can't withstand these most basic loads.

The communications around this have indeed been a mess.

The driver of the outages is indeed what PG&E claims, but it's not necessarily in your area. Your city may be affected simply because they need to de-energize transmission lines (that run through a high-risk area) that serve your city.

I had the same thought. But when the power was turned off in my town last night, the corridor that was affected was not near us. PG&E should have known that (1) the conditions were not even close to what they had thought on the ground, and (2) except for certain transmission corridors in our area, no one would be affected. Instead they caused massive inconvenience and minor panic, with businesses closing and schools spending lots of time planning for unnecessary contingencies.

Reply on news.ycombinator.com