KVCache simulator for LLM serving cluster routing research

Discrete-event simulator for evaluating KV cache-aware routing policies
in prefill-disaggregated LLM serving clusters. Models a two-tier KV cache
hierarchy (L0 GPU HBM + L1 CPU DRAM) with RDMA/PCIe link contention,
architecture-derived roofline compute (MoE, MLA, DSA), and a cluster-wide
meta-store for prefix-aware routing decisions.

Includes 11 routing policies (random, round_robin, least_loaded,
least_tokens, ttl_aware, precise, min_pd, cache_load, cache_score,
estimated_ttft, prefix_affinity), HuggingFace config.json auto-parsing,
built-in GPU hardware presets (H100/H800/H20/A100/B200), and ablation
tooling for systematic policy comparison across real Alibaba serving traces.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

src/metrics/routing_log.rs

@@ -0,0 +1,29 @@
+use anyhow::Result;
+use std::fs::File;
+use std::io::{BufWriter, Write};
+use std::path::Path;
+
+use crate::router::RouteDecision;
+
+pub struct RoutingLogWriter {
+    inner: BufWriter<File>,
+}
+
+impl RoutingLogWriter {
+    pub fn create<P: AsRef<Path>>(path: P) -> Result<Self> {
+        let f = File::create(path)?;
+        Ok(Self { inner: BufWriter::new(f) })
+    }
+
+    pub fn write(&mut self, decision: &RouteDecision) -> Result<()> {
+        let line = serde_json::to_string(decision)?;
+        self.inner.write_all(line.as_bytes())?;
+        self.inner.write_all(b"\n")?;
+        Ok(())
+    }
+
+    pub fn finish(mut self) -> Result<()> {
+        self.inner.flush()?;
+        Ok(())
+    }
+}