[feature] [sgl-router] Add a dp-aware routing strategy (#6869)

This commit is contained in:
Rui Chen
2025-07-30 20:58:48 +08:00
committed by GitHub
parent 55ecdc0a8e
commit a730ce8162
19 changed files with 726 additions and 16 deletions

View File

@@ -37,6 +37,8 @@ struct Router {
eviction_interval_secs: u64,
max_tree_size: usize,
max_payload_size: usize,
dp_aware: bool,
api_key: Option<String>,
log_dir: Option<String>,
log_level: Option<String>,
service_discovery: bool,
@@ -136,6 +138,8 @@ impl Router {
request_timeout_secs: self.request_timeout_secs,
worker_startup_timeout_secs: self.worker_startup_timeout_secs,
worker_startup_check_interval_secs: self.worker_startup_check_interval,
dp_aware: self.dp_aware,
api_key: self.api_key.clone(),
discovery,
metrics,
log_dir: self.log_dir.clone(),
@@ -161,6 +165,8 @@ impl Router {
eviction_interval_secs = 60,
max_tree_size = 2usize.pow(24),
max_payload_size = 256 * 1024 * 1024, // 256MB default for large batches
dp_aware = false,
api_key = None,
log_dir = None,
log_level = None,
service_discovery = false,
@@ -193,6 +199,8 @@ impl Router {
eviction_interval_secs: u64,
max_tree_size: usize,
max_payload_size: usize,
dp_aware: bool,
api_key: Option<String>,
log_dir: Option<String>,
log_level: Option<String>,
service_discovery: bool,
@@ -225,6 +233,8 @@ impl Router {
eviction_interval_secs,
max_tree_size,
max_payload_size,
dp_aware,
api_key,
log_dir,
log_level,
service_discovery,