Openflow questions

[Alba]

Hello everyone,

We want to ask you two questions that arise about Openflow:

1.- Is it possible to perform load balancing between the different robots?

2.- Is there a maximum limit of robots that can connect to Openflow?

Greetings and thanks for everything.

[Allan_Zimmermann]

answering in reverse.

• 2 OpenFlow looks at clients; it does not care what type the client is (web, robot, agent, PowerShell, etc.). There is NO software limit in OpenFlow, but the resources on the server(s) being used will set limits. Depending on the type of work the clients do, if they perform a lot of database operations, you may observe high RAM usage when parsing results and/or issues with the database keeping up. That’s why I always recommend using MongoDB Atlas for production or high-load systems.

• 1 Yes, there are multiple ways to achieve that. I’ll list the two most common methods here:

• • a) The most basic and simple way is by using roles. On a role, you can enable “RPA.” This will instruct all robot users who are members of that role to listen for messages sent to a queue associated with that role. You can then send requests to run workflows to the role instead of individual robots. This approach is suitable for ad hoc tasks, but for most scenarios, option b is much better.
• • b) Another approach is to use work item queues. By dividing your work into smaller units (or splitting the processing into multiple steps across multiple work item queues), you can easily add or remove robots from your work queues. This method still utilizes the same role as mentioned above, but instead of sending a message to a message queue, you add work to the work item queue. OpenFlow handles failures, load balancing, and provides access to ways to report on status.

Workitems

[Allan_Zimmermann]

small add to the above.
In my unit tests i wrote a small load tester, that can add any number of clients and then make them do random database actions, to simulate work ( this will NEVER be like real life, but still )
on the smallest server you can get in google clouds, this server could handle more than 2500 without any issues ( at 3000 I had issue, but was not sure if that was in my client code or due to the server not being able to keep up. But I’m still happy with 2500 )

Allan Zimmermann on LinkedIn: #rpa #openrpa #openflow

[Alba]

Thanks you very much for your response.

When we have all the robots available we check it.

Best Regards.