3 ms·
It's not the nicest thing in the world, but his university (UNC, as far as I can tell) has 30,000 students. 130 requests/min is the equivalent of having 0.4% of
by harpastum 17y ago
It's not the nicest thing in the world, but his university (UNC, as far as I can tell) has 30,000 students. 130 requests/min is the equivalent of having 0.4% of the students on at one time.
It obviously creates a visible imprint, but it shouldn't be a show-stopper for the school.
- zaidf 17y agoYeah and like I said, it didn't cause a problem for over 3 weeks. And I am still not convinced it was the culprit in the last few days. From what I've experienced and from friends who went here five years ago, their system goes down/has issues every year during end of semester. This year they seem to have a scapegoat to blame it all on...except it doesn't help to admit that they are still having issues even after blocking me.
- jerf 17y agoDevil's advocate: Without knowing the details, I would observe that the load generated by searches can be very different. Searching for all 100-level Physics classes may be an optimized code path (or DB query) that occurs in a fraction of a second, where as "Give me all classes in a department starting with P" might be slow, or "Give me all courses taught by Prof. Bill" might go down an unindexed code path. They might never have noticed if normal users only fired it only once per hour, but if you happen to be using more of those queries than a normal student because it's more convenient for you, then you might actually be hammering their system without meaning to. Or, put another way, "queries per second" is a terrible metric for "how much load I'm subjecting their system to". :) I can give you one bound you might want to instrument up to check their contention about load: How quickly they answer. If they answer in 1ms (minus ping time), then there's a sharp bound on how much of their resources you can be taking up. (I'm going to assume bandwidth isn't much of a concern here, based on the numbers.) If you see some queries taking longer, perhaps even into the seconds, then they may have a point. It could just be general load, or it could be your thing specifically.