Why the performance of the servers in the cluster below?

Set up on the basis of copies of the original web server of the web cluster Bitrix instructions. 4 VDS servers: mgmt+web, db master, db slave, memcache. Each server is 8 GB + 4x2,4 GHz + SSD (TimeWeb). Timings between servers minimum, all servers are tuned for maximum performance, the storage of temporary tables in the database not in memory. Panel performance (see image 1) gives no comforting results.
On the original server where all services in one place, the results were much better (see picture 2).
What could be the reason for such bad results?
March 19th 20 at 08:37
2 answers
March 19th 20 at 08:39
Clustering is still required to improve stability, not "speed". Introducing, for example, the database on a third server you have much to lose in speed, it can be seen from your screenshots.

Roughly speaking, the fastest configuration is when all services are installed on 1 server.
March 19th 20 at 08:41
Check the configs database.
In the manual database is not located on the servers of statics - so you can ONLY do if you have a large project and a pool of database servers.
Ideally, the Mysql master should be on the balancer to set - otherwise all queries start walking on the network with appropriate capacity.

Find more questions by tags Replication1C-BitrixClustering