How to optimize resource consumption by the 1C server?

Question

Server optimization

The point is this: I am an admin, and I am familiar with 1C only from the database backup side, i.e. practically unaware of the inner workings of 1C.
On the fingers, the scheme is as follows: Windwos Server 2012 x64 is used everywhere. The sql server is running on one virtual machine, the 1c server is running on the second, and client sessions are started on the third using RemoteApps.
When everything was spinning on version 1C 8.2. **** there were no hard lags. The consumption of processor power of the hypervisor was divided approximately as follows: 4-7% for the 1c server, 7-9% for sql and 15-20% for the terminal client server. RAM was given to everyone for the eyes - 32 GB each
Then the programmers rolled up the update to 8.3 and the load on the 1c server increased to 50% of the physical processor power. Users swear, it’s impossible to work, but it’s impossible to get anything intelligible from programmers, they say, go google, we don’t code shit.
So, the question is: what thin places can be in such a configuration: a database of about 35 GB on MS SQL Server 2008r2, server version 1c 8.3.7.1790. All on x64. Number of users - 30.

Answer 1

Which process is eating up resources? If rphost - then this is the base (and configuration - it means the code of the curve), if rmngr - then the services (it means the environment is crooked).
Skul and the server are better placed on the same wheelbarrow - then the work will not be over the network, this will greatly speed up.
In the cluster, in the properties of the working server, we put a checkbox - Manager for each service - we look at which of the rpmng eats resources, by its process PID we find which service it is in the cluster - we draw conclusions.
Custom config? Debugging needs to be started, to understand which request heavily loads resources.

Answer 2

1. 8.3.7.17xx, in principle, is not the most successful platform release branch to compare with 8.2, but if we forget about platform errors, then the most important thing is that in 8.3 the 1C server part was completely redesigned, including files with session data are now compressed, which from the point of view of space it is better, but the load on the percentage is higher, and there are quite a lot of such features
2. statistically, the load on the hard disk is higher, and you can try switching the log file from the sqllite format to the old mode, and processor resources are needed much more than in 8.2, so do not think that hardware suitable for 8.2 will remain so for 8.3 - request a budget and increase the capacity for the 8.3 server