G
George Potemkin
Guest
> the current PROMON, "debghb", 11 has two columns on the far right "nap max total / hwm" that seem like they might be good candidates to add to the _Latch VST. My thinking there is that they would give better insight into how frequently the -spin limit is being reached, how much napping is actually taking place and how long those naps are. They are useful only when -napmax is larger than -nap. But why they are set so? Normal duration of latch lock is less than 100 nanoseconds. Default -nap is 10 milliseconds. In other words, if a process failed to get a latch it kindly allows other processes to lock the latch almost 100,000 times (= 10 ms / 100 ns). The higher nap time will increase the "kindness" of our process but does its job is less important than of other processes? This only makes a responce time less predictable. Of course, the -napmax was introduced to save CPU time: spinning consumes CPU but does not make any useful changes in database. The "courtesy" of our process allows other processes to consume CPU by spinning when a latch is busy. When a resource is a bottleneck then law of the jungle comes on a stage: -napmax = -nap = 1 ms.
Continue reading...
Continue reading...