Weekend thought experiment? Cannot download a WU first time, second time, never? The techs wrote in prepping for OPN1 launch they were scaled up for an increase by a factor 8 or something, so 'any' means from 2 million to 16 million daily I suppose.

In the limit:

Say the number is X. It would mean that if in the next few years crunchers would on average invite X as enthusiastic crunchers as them, researchers would bearly wait (I guess only the runtime of the heaviest WU) after placing a sub-project online before getting all its output. In other words, it would be the highest computing capacity that WCGrid would ever need.

Yes, I understand this is utopic, but it could give us a little bit of insight into where we are now.

By the way, don't think of it as a thought experiment, rather a reality check.

You Folks have too much time on your hands.... Go pick up some cheap used systems with obsolete parts that are hard to find. Start running more WCG WU's until they break the systems and then occupy yourself constructively fixing your Geezer systems. That will keep you busy enough to not be asking / answering questions like this.

Bill F

There's nothing wrong with having a discussion. You have no right to judge how people use their free time.

@wcgridmember, seems like the answer to your original question is basically the same as looking for the weakest link in WCG's IT systems. aka "Find the bottleneck."

In other words, you're simply asking for the breaking point in WCG's ability to distribute work units, so it makes sense to take a look at each link in the chain of work unit distribution. From this recent thread, armstrdj says:

So we'd need more details on how much capacity WCG has in their server farm that runs Apache Mesos and Aurora as well as any load balancers, routers, and switches in their infrastructure. I think they moved to IBM Cloud a few years ago.

Overload any of those links in the chain, and it would prevent members from downloading more work units until the overcapacity is fixed.

To really answer your question, we'd have to measure the percent capacity for each of those links by maybe looking at system logs for the last several weeks/months and comparing it to the maximum speed or capacity.
---------------------
Or maybe your original question isn't related to bottlenecks at all. Maybe you were asking how many more factors of the current active members would be needed to complete all projects immediately (without taking into account infrastructure bottlenecks and assuming unlimited capacity)? As in, "Let's throw in 10 million more volunteers at the same time and assume their IT infrastructure doesn't break....does that crunch through all the work in a day?"