This content has been marked as final. Show 3 replies
I'm afraid that the short answer is that you can't.
There are variosu possibilities but all have their disadvantages. If you really (a) cannot open multipel ports in the firewall and (b0 have bandwidth/latency limitations on the link between the datacenters then my best suggestion is to go with your current setup as far as SUBSCRIBER (T15) and write a custom XLA program that runs on T15, captures all the changes that arrive there via replication from T6 and either (a) does it's own propagation to the other datastores in DS2 or (b) use T15 as purely the receiving point for replication from DS1 and the XLA program then inserts into one of a set of replicated TT in second DS.
|--- XLA --> SUBSCRIBER (T1)
MASTERS(T1,...) --- TT rep --> PROPAGATOR(T6) ---------- TT rep -----------> SUBSCRIBER(T15) -- XLA ---> SUBSCRIBER(T2)
|--- XLA --> SUBSCRIBER (T3)
|----- TT rep --> SUBSCRIBER (T2)
MASTERS(T1,...) --- TT rep --> PROPAGATOR(T6) ---------- TT rep -----------> SUBSCRIBER(T15) -- XLA ---> SUBSCRIBERS(T1) --- TT rep --> SUBSCRIBER(T3)
|----- TT rep --> SUBSCRIBER (T4)
I'm not sure my diagrams are very good but hopefully you get the idea.
Thanks Chris. I was thinking of the same solution.
One thing though is not clear to me is as how we can have HA in non active-standby replication configuration.
I don't quite understand the question. Can you please explain in more detail and in particular clarify what you mean by High Availability in this context.