Unfortunately, we still could not solve the production-preventing problem. Various activities, such as Switch VAAI off and on, update HBA driver according to VMware recommendation, ACT Heartbeat off and on, etc. have so far unfortunately not brought any success.
The last piece of knowledge we've learned is that if you recompose to two disks in the View 7.4 / ESX 6.5 environment, one LUN gets a high write rate (about 120mb / s), while the second LUN gets 13mb / s dulls itself and this value also does not increase when the performant LUN has finished writing.
We have now determined comparative values in the "old" environment under View 6.2.3 / ESX 5.5 and were able to detect a parallel write rate of approx. 50 mb / s in two recorders at the same time. Therefore, the replica files are also finished relatively at the same time and we do not get a timeout.
Can you explain the phenomenon of different writing rates? Can it be that the View Composer is responsible for this and there is the problem to look for? The assumption comes from the fact that VAAI is used in a clone of a machine and here View is out, whereas in a recompose no VAAI is used. Why?
Sorry for my bed english
