Hi Mike, probably so, but due to large amount of participants can't confirm it. Is there a way to identify this slowdown to confirm the theory? Is there a way to avoid slowdown, like setting the baseline for the performance for the shared connection? Or is there a way to identify a slow client, so maybe we'll be able to drop it in favor of performance for the majority?
On Thu, 9 Feb 2023, 21:08 Michael Jumper, <[email protected]> wrote: > Did anyone on the connection have a very slow network connection? The > server may have needed to reduce its responsiveness to avoid streaming far > more data than that user would be able to receive and process in the same > amount of time it took the server to generate that data. > > Network latency shouldn't be an issue, as the Guacamole server is capable > of distinguishing between client slowness due to network speed vs. network > latency, and makes constant adjustments accordingly. > > There should not be any appreciable increase in processing load for shared > connections, as shared connections in Guacamole work by sending the same > data to all users of the connection. The more intensive part of the > processing involved in a connection, graphical operations and compression, > are performed only once. > > If the presenter was experiencing slowness in the movement of the local > mouse cursor, that suggests an issue with the local machine and not the > shared connection. The mouse cursor is rendered locally, with few > exceptions. > > - Mike > > On Thu, Feb 9, 2023, 4:33 AM Vladimir Sorokin <[email protected]> wrote: > >> Hi, >> We encountered performance issues for shared RDP connection with 150+ >> users on the same connection. >> We have a custom guacamole application that consists of multiple docker >> glyptodon/guacd:2.8.1 (we intend to align and use latest >> guacamole/guacd:1.4), multiple guacamole-common:1.4 proxy servers, and >> guacamole-client:1.2. >> We have a scenario with a shared RDP connection, where only one user >> (presenter) has write privileges and others only read only. >> We've been using this setup for years without a problem with shared >> connections with up to 100 users, and only recently have we encountered >> performance problems for shared connection with 150+ users. >> We monitored the underlying AWS ec2 machines' resource usage (CPU, >> memory, network), and it does not appear that the resource usage is out of >> the baseline, so it does not look like the problem originates in the >> resource usage. >> On the shared connection, the user with the write privileges presents >> some desktop application, nothing graphical, CPU, or network heavy. The >> network connection for the presenter seems stable. >> And yet, the presenter experiences significant lag on every action - >> clicks, cursor movement, etc. >> Could there be a performance limit for the maximum number of users per >> shared connection? >> We intend to execute a load test to confirm, but maybe there are other >> directions that we should look at? >> >> Thanks >> >> -- >> Vladimir Sorokin >> Developer >> +972 54 723 6151 >> >
