Decide what the default send and receive buffer sizes should be and document the reasoning behind that choice. The goal is to answer questions about latency, throughput, and memory tradeoffs as a function of the send/receive buffer size.
The current default is 0 for both (synchronized channel). We want to measure their behavior under a few representative workloads and compare them with a few alternative settings. The outcome should be a concrete decision, not just raw benchmark output. If the current defaults are acceptable, the task can conclude by documenting why they stay as-is.
The cmd/benchmark tool has knobs for tuning these buffer sizes. We should evaluate these questions for both the WAN and LAN cluster settings.
Decide what the default send and receive buffer sizes should be and document the reasoning behind that choice. The goal is to answer questions about latency, throughput, and memory tradeoffs as a function of the send/receive buffer size.
The current default is 0 for both (synchronized channel). We want to measure their behavior under a few representative workloads and compare them with a few alternative settings. The outcome should be a concrete decision, not just raw benchmark output. If the current defaults are acceptable, the task can conclude by documenting why they stay as-is.
The
cmd/benchmarktool has knobs for tuning these buffer sizes. We should evaluate these questions for both the WAN and LAN cluster settings.