Yes, this can work, and will achieve my stated goal. My stated goal, however, does not tell the whole story...
I suppose my follow-up to this would be -- Can BIRD run in a shadow/read-only mode, allowing a primary BIRD process to read/write data into memory, but multiple BIRD processes to read the primary's memory space, each with a unique control socket for direct queries?
The above is meant to serve as an extremely scalable system for querying BIRD directly, while distributing queries across multiple read-only workers; at the same time not duplicating the memory space of the parent BIRD process for each worker. As it stands, the BIRD memory space I am trying to query consumes about 1.1GB, and yields 5,000 queries per second before a core reaches 99% CPU. I would like to scale this out, and achieve (5,000 * n) QPS without also consuming (1.1GB * n) memory.
Thoughts?