Add vLLM dashboard and integration (#882)

Adds a vLLM sample dashboard, associated with integration metadata (currently HIDDEN; will launch after validation and docs updates). go/vllm-dashboard-dev-notes (internal link)
GoogleCloudPlatform · Dec 16, 2024 · 5ffe87e · 5ffe87e
1 parent e492dfa
commit 5ffe87e
Show file tree

Hide file tree

Showing 7 changed files with 1,600 additions and 0 deletions.
diff --git a/dashboards/vllm/README.md b/dashboards/vllm/README.md
@@ -0,0 +1,10 @@
+### Dashboards for Velero
+
+#### Notes
+
+- This dashboard is based on built-in prometheus metrics.
+
+|vLLM Prometheus Overview|
+|:------------------|
+|Filename: [vllm-prometheus.json](vllm-prometheus.json)|
+|This dashboard has charts displaying throughput, latency, cache usage and other metrics for vLLM inference.|
diff --git a/dashboards/vllm/metadata.yaml b/dashboards/vllm/metadata.yaml
@@ -0,0 +1,9 @@
+sample_dashboards:
+  -
+    category: vLLM
+    id: vllm-prometheus
+    display_name: vLLM Prometheus Overview
+    description: "This dashboard has charts displaying throughput, latency, cache usage and other metrics for vLLM inference."
+    related_integrations:
+      - id: vllm
+        platform: GKE
diff --git a/dashboards/vllm/vllm-prometheus.01.png b/dashboards/vllm/vllm-prometheus.01.png
diff --git a/dashboards/vllm/vllm-prometheus.02.png b/dashboards/vllm/vllm-prometheus.02.png