In Servoy 5 there is server clustering.
How about the batch processors. Does Servoy decide on what server it runs, or can the batch processor be attached to one of the servers.
In that way I can separate (process consuming) batch processing from standard online client processing.
Batch processors are specified in servoy.properties file. You have separate servoy.properties files for each server in the cluster => you can specify which batch processors run on each server .
Andrei,
Can I also make configuration that certain solutions can only run on 1 specific cluster.
I want 1 server to use for permanent batch processing and a second server for normal client access.
I ask this because those headless batch processors take lot of CPU and I don’t want the standard clients run on the same server.
Martin
martinh:
Can I also make configuration that certain solutions can only run on 1 specific cluster.
You mean server, not cluster right?
You can start a cluster containing 2 servers, only one of them running batch processors, and only the other one being visible to client access. (you only set up a Servoy Server Cluster without load-balancing in this case and hide the batch-processor server (firewall/router)).
You should also think about what the batch processors are doing - if for example they do heavy data update operations, data broadcasts will possibly still reach the main (client accessible) server using some of the CPU - if changed data will affect solution running on main server. If you do only heavy CPU processing and tasks that don’t require too much traffic between servers that would be ideal.
Andrei Costescu:
martinh:
Can I also make configuration that certain solutions can only run on 1 specific cluster.You mean server, not cluster right?
Yes of course
Andrei Costescu:
You should also think about what the batch processors are doing - if for example they do heavy data update operations, data broadcasts will possibly still reach the main (client accessible) server using some of the CPU - if changed data will affect solution running on main server. If you do only heavy CPU processing and tasks that don’t require too much traffic between servers that would be ideal.
These batch processors will be doing lot of database updates. It will process interface files from/to our old application. So there are lot of DB-updates.
I noticed that when such an update is running, sometimes smart client can’t even make connection to the server, because that processing is so busy
You should see an improvement anyway as only data-broadcasts would reach the main client server.
I guess that a lot of parsing/other operations that happen before the updates will be moved to the other server completely.