In Grizzly-based HTTP connectors, how are selector pools allocated for Listeners and Requesters?

In Grizzly-based HTTP connectors, how are selector pools allocated for Listeners and Requesters?

• A. Listeners and Requesters share a single pool
• B. Listeners use a shared pool per app; Requesters use a dedicated pool per app
• C. Requesters share a pool per app; Listeners have a dedicated pool per app
• D. Each listener/requester has its own global pool

Answer: (B)

Selector pools handle the threads that drive non-blocking I/O for Grizzly-based HTTP connectors. For listeners, sharing a single pool per application makes sense because multiple server endpoints within the same app can efficiently multiplex on a common set of selector threads, reducing overhead and improving resource utilization. For requesters, using a dedicated pool per application provides isolation: all outbound calls from that app run on their own pool, so heavy outbound activity in one app doesn’t impact other apps and tuning can be tailored to outbound behavior (timeouts, TCP connections, TLS handshakes) without affecting listeners. The design favors efficiency for handling incoming connections while preserving predictable, isolated performance for outbound requests.

Selector pools handle the threads that drive non-blocking I/O for Grizzly-based HTTP connectors. For listeners, sharing a single pool per application makes sense because multiple server endpoints within the same app can efficiently multiplex on a common set of selector threads, reducing overhead and improving resource utilization. For requesters, using a dedicated pool per application provides isolation: all outbound calls from that app run on their own pool, so heavy outbound activity in one app doesn’t impact other apps and tuning can be tailored to outbound behavior (timeouts, TCP connections, TLS handshakes) without affecting listeners. The design favors efficiency for handling incoming connections while preserving predictable, isolated performance for outbound requests.