It won’t. I was thinking more about backend components where you allocate a fixed per request memory budget. It would work best if you could apply the budget pe

It won’t. I was thinking more about backend components where you allocate a fixed per request memory budget. It would work best if you could apply the budget per in-flight request. In case a request goes over budget, only the affected (presumably nefarious) request gets terminated, other requests and other processes will run uninterrupted.