Background:

1. wt does not have average time consuming statistics. Adding total time consuming statistics can help us obtain average time consuming of nodes. Average time consuming =perf_hist_opwrite_latency_total/(perf_hist_opwrite_latency_lt100+perf_hist_opwrite_latency_lt250+ perf_HIST_opwrite_opency_LTWrite_ latency_lt500+perf_hist_opwrite_latency_lt1000+perf_hist_opwrite_latency_lt10000+perf_hist_opwrite_latency_gt10000)/ Sampling period.
2. Sometimes we often encounter node delay jitter, and we do not know whether it is caused by mongo server or wiredtiger. Increasing the statistics can help quickly locate the problem caused by who
3. by using (perf_hist_opwrite_latency_lt100+perf_hist_opwrite_latency_lt250+perf_hist_opwrite_latency_lt500+perf_hist_opwrite_laten cy_lt1000+perf_hist_opwrite_latency_lt10000+perf_hist_opwrite_latency_gt10000)/( Sampling period ) , the read/write qps of nodes can be obtained, which can help us obtain qps statistics.