To be fair, even our Australian unemployment system crashed under the load last week. It was designed to handle up to 50,000 users at a single time - which, under normal circumstances, is...
Now, as thousands of people try to get help, the system crashes or denies them access. Nearly 400,000 people have managed to file claims in the last two and half weeks. It’s not known how many have tried and failed.
To be fair, even our Australian unemployment system crashed under the load last week. It was designed to handle up to 50,000 users at a single time - which, under normal circumstances, is unrealistically high. There's never ever ever going to be a time when 50,000 people decide to access the unemployment system at the same time.
Until the government shuts down all non-essential services overnight and a million people lose their jobs in a single week, and 500,000 newly unemployed people try to sign up in one day for the new & enhanced unemployment benefits deliberately aimed at supporting those hundreds of thousands of people.
The Minister for Government Services stood up in Parliament (which happened to be sitting that day) and told everyone that the unemployment system was suffering a DDoS attack, which is why it had shut down automatically. An hour later, he had to change his story in a press conference: it seems that there were so many legitimate attempts to access the website that it looked like a DDoS attack.
While I can't speak specifically to Florida's circumstances, it's fair to assume that even a properly designed system wouldn't handle the extreme and unexpected loads caused by the current circumstances.
I can forgive the technical shortcomings like what you're describing. Looking at historical spikes and then adding some wiggle room doesn't even get you close to the numbers we're seeing. This is...
I can forgive the technical shortcomings like what you're describing. Looking at historical spikes and then adding some wiggle room doesn't even get you close to the numbers we're seeing.
This is the part that irks me:
“It wasn’t about saving money. It was about making it harder for people to get benefits or keep benefits so that the unemployment numbers were low to give the governor something to brag about.”
I think that depends on what you mean by "properly designed." Many systems do fall over when they get 10x the load they're used to. But then again, this isn't a lot of traffic compared to popular...
I think that depends on what you mean by "properly designed." Many systems do fall over when they get 10x the load they're used to. But then again, this isn't a lot of traffic compared to popular Internet services, and I would expect a properly designed server app that's running in the cloud to scale up pretty well since it's not like they are doing any heavy calculations.
Also "properly tested". Even if you had an experienced team build it, one innocuous mistake or oversight can cripple an entire app when it comes under heavy load. When they had a conversation with...
Also "properly tested". Even if you had an experienced team build it, one innocuous mistake or oversight can cripple an entire app when it comes under heavy load. When they had a conversation with management about how they want to do load testing to see if will handle X users, they managers probably laughed about what a ridiculous number it was and said something to the effect of "that'll never happen, and it's running fine now, why would we do that?".
Privately, Republicans admit that the $77.9 million system that is now failing Florida workers is doing exactly what Scott designed it to do — lower the state’s reported number of jobless claims after the great recession.
[...]
With hundreds of thousands of Floridians out of work, the state’s overwhelmed system is making it nearly impossible for many people to even get in line for benefits.
After a record number of claims were reported Thursday, DeSantis said the state would resort to paper applications, build a mobile app to handle the flood of traffic and deploy hundreds, even thousands, of state workers to provide stopgap help.
To be fair, even our Australian unemployment system crashed under the load last week. It was designed to handle up to 50,000 users at a single time - which, under normal circumstances, is unrealistically high. There's never ever ever going to be a time when 50,000 people decide to access the unemployment system at the same time.
Until the government shuts down all non-essential services overnight and a million people lose their jobs in a single week, and 500,000 newly unemployed people try to sign up in one day for the new & enhanced unemployment benefits deliberately aimed at supporting those hundreds of thousands of people.
The Minister for Government Services stood up in Parliament (which happened to be sitting that day) and told everyone that the unemployment system was suffering a DDoS attack, which is why it had shut down automatically. An hour later, he had to change his story in a press conference: it seems that there were so many legitimate attempts to access the website that it looked like a DDoS attack.
While I can't speak specifically to Florida's circumstances, it's fair to assume that even a properly designed system wouldn't handle the extreme and unexpected loads caused by the current circumstances.
I can forgive the technical shortcomings like what you're describing. Looking at historical spikes and then adding some wiggle room doesn't even get you close to the numbers we're seeing.
This is the part that irks me:
I think that depends on what you mean by "properly designed." Many systems do fall over when they get 10x the load they're used to. But then again, this isn't a lot of traffic compared to popular Internet services, and I would expect a properly designed server app that's running in the cloud to scale up pretty well since it's not like they are doing any heavy calculations.
Also "properly tested". Even if you had an experienced team build it, one innocuous mistake or oversight can cripple an entire app when it comes under heavy load. When they had a conversation with management about how they want to do load testing to see if will handle X users, they managers probably laughed about what a ridiculous number it was and said something to the effect of "that'll never happen, and it's running fine now, why would we do that?".
From the article:
[...]