SQL Anywhere 17 Read-only scale-out: how does the root server choose copy node?

[Arthoor]

I am setting up SQL Anywhere 17 Read-only scale-out system. I have a question about this sentence in the docs: "You can choose to have the application connect to the copy node that is least heavily loaded using the NodeType=COPY connection parameter".

I would like to know as precisely as possible how the root server chooses which copy node to connect to, i. e. how does the root server know which copy node is least heavily loaded at that moment?

Thanks.

[VolkerBarth · Accepted solution]

While I do not know how exactly the load on different copy nodes is calculated, I'm quite sure the system procedure sa_mirror-server_status() does provide the results (including several "load" values for current state vs. last n minutes) and helps the Read-only Scale-out facility to choose the node to connect to. According to the linked doc, that decision is taken:

"When the NodeType connection parameter is specified, the database server uses load information to redirect connections. The database server selects the mirror server with the lowest load; if all servers have the same load, the server with the fewest connections is used."

---

[Arthoor · Accepted solution]

That's basically what I wanted to know. Thanks.