[Serve][Doc] Add http endpoint for dag pattern doc (#25243)

doc/source/_toc.yml

@@ -139,6 +139,7 @@ parts:
                   - file: serve/deployment-graph/chain_nodes_same_class_different_args
                   - file: serve/deployment-graph/combine_two_nodes_with_passing_input_parallel
                   - file: serve/deployment-graph/control_flow_based_on_user_inputs
+                  - file: serve/deployment-graph/http_endpoint_for_dag_graph
           - file: serve/architecture
           - file: serve/tutorials/index
           - file: serve/faq

doc/source/serve/deployment-graph.md

@@ -13,4 +13,5 @@ Jump striaght into a common design patterns using deployment graph:
 - [Chain nodes with same class and different args](deployment-graph/chain_nodes_same_class_different_args.md)
 - [Combine two nodes with passing same input in parallel](deployment-graph/combine_two_nodes_with_passing_input_parallel.md)
 - [Control flow based on user inputs](deployment-graph/control_flow_based_on_user_inputs.md)
+- [Http endpoint for dag graph](deployment-graph/http_endpoint_for_dag_graph.md)
 

doc/source/serve/deployment-graph/control_flow_based_on_user_inputs.md

@@ -1,6 +1,6 @@
 # Pattern: Control flow based on the user inputs
 
-The example shows how to user inputs to control the graph flow
+The example shows how to use inputs to control the graph flow
 
 ![pic](https://raw.githubusercontent.com/ray-project/images/master/docs/serve/deployment-graph/control_flow_based_on_user_inputs.svg)
 

doc/source/serve/deployment-graph/http_endpoint_for_dag_graph.md

@@ -0,0 +1,29 @@
+# Pattern: Http endpoint for dag graph
+
+This example shows how to configure ingress component of the deployment graph, such as HTTP endpoint prefix, HTTP to python object input adapter.
+
+## Code
+
++++
+
+```{eval-rst}
+.. literalinclude:: ../doc_code/deployment_graph_dag_http.py
+   :language: python
+```
+
+````{note}
+1. Serve provide a special driver ([DAGDriver](deployment-graph-e2e-tutorial.html#step-4-driver-deployment-to-handle-http-ingress)) to accept the http request and drive the dag graph execution
+2. User can specify the customized http adapter to adopt the cusomized input format
+````
+
+## Outputs
+
+Model output1: 1(input) + 0(weight) + 4(len("test")) = 5 \
+Model output2: 1(input) + 1(weight) + 4(len("test")) = 6 \
+So the combine sum is 11
+```
+11
+11
+```
+
++++

doc/source/serve/doc_code/deployment_graph_dag_http.py

@@ -0,0 +1,54 @@
+import requests
+from pydantic import BaseModel
+
+import ray
+from ray import serve
+from ray.serve.drivers import DAGDriver
+from ray.experimental.dag.input_node import InputNode
+
+
+ray.init()
+serve.start()
+
+
+class ModelInputData(BaseModel):
+    model_input1: int
+    model_input2: str
+
+
+@serve.deployment
+class Model:
+    def __init__(self, weight):
+        self.weight = weight
+
+    def forward(self, input: ModelInputData):
+        return input.model_input1 + len(input.model_input2) + self.weight
+
+
+@serve.deployment
+def combine(value_refs):
+    return sum(ray.get(value_refs))
+
+
+with InputNode() as user_input:
+    model1 = Model.bind(0)
+    model2 = Model.bind(1)
+    output1 = model1.forward.bind(user_input)
+    output2 = model2.forward.bind(user_input)
+    dag = combine.bind([output1, output2])
+    serve_dag = DAGDriver.options(route_prefix="/my-dag").bind(
+        dag, http_adapter=ModelInputData
+    )
+
+dag_handle = serve.run(serve_dag)
+
+print(
+    ray.get(
+        dag_handle.predict.remote(ModelInputData(model_input1=1, model_input2="test"))
+    )
+)
+print(
+    requests.post(
+        "http://127.0.0.1:8000/my-dag", json={"model_input1": 1, "model_input2": "test"}
+    ).text
+)