PIPELINE_SERVING.md

# Pipeline Serving

([简体中文](PIPELINE_SERVING_CN.md)|English)


Paddle Serving is usually used for the deployment of single model, but the end-to-end deep learning model can not solve all the problems at present. Usually, it is necessary to use multiple deep learning models to solve practical problems.

Paddle Serving provides a user-friendly programming framework for multi-model composite services, Pipeline Serving, which aims to reduce the threshold of programming, improve resource utilization (especially GPU), and improve the prediction efficiency.

## ★ Architecture Design

The Server side is built based on <b>RPC Service</b> and <b>graph execution engine</b>. The relationship between them is shown in the following figure.

<center>
<img src='pipeline_serving-image1.png' height = "250" align="middle"/>
</center>

### 1. RPC Service

In order to meet the needs of different users, the RPC service starts one Web server and one RPC server at the same time, and can process 2 types of requests, RESTful API and gRPC.The gPRC gateway receives RESTful API requests and forwards requests to the gRPC server through the reverse proxy server; gRPC requests are received by the gRPC server, so the two types of requests are processed by the gRPC Service in a unified manner to ensure that the processing logic is consistent.

#### <b>1.1 Request and Respose of proto

gRPC service and gRPC gateway service are generated with service.proto.

```proto
message Request {
  repeated string key = 1;  
  repeated string value = 2;
  optional string name = 3;
  optional string method = 4;
  optional int64 logid = 5;
  optional string clientip = 6;
};

message Response {
  optional int32 err_no = 1;
  optional string err_msg = 2;
  repeated string key = 3;
  repeated string value = 4;
};
```

The `key` and `value` in the Request are paired string arrays. The `name` and `method` correspond to the URL of the RESTful API://{ip}:{port}/{name}/{method}.The `logid` and `clientip` are convenient for users to connect service-level requests and customize strategies.

In Response, `err_no` and `err_msg` express the correctness and error information of the processing result, and `key` and `value` are the returned results.

### 2. Graph Execution Engine

The graph execution engine consists of OPs and Channels, and the connected OPs share one Channel.

- Channel can be understood as a buffer queue. Each OP accepts only one Channel input and multiply Channel outputs (each output is the same); a Channel can contain outputs from multiple OPs, and data from the same Channel can be used as input for multiple OPs.
- Users only need to define relationships between OPs. Graph engine will analyze the dependencies of the entire graph and declaring Channels at the compile time.
- After Request data enters the graph execution engine service, the graph engine will generator an Request ID, and Reponse is returned through corresponding Request ID.
- For cases where large data needs to be transferred between OPs, consider RAM DB external memory for global storage and data transfer by passing index keys in Channel.

<center>
<img src='pipeline_serving-image2.png' height = "300" align="middle"/>
</center>


#### <b>2.1 OP Design</b>

- The default function of a single OP is to access a single Paddle Serving Service based on the input Channel data and put the result into the output Channel.
- OP supports user customization, including preprocess, process, postprocess functions that can be inherited and implemented by the user.
- OP can set the number of concurrencies to increase the number of concurrencies processed.
- OP can obtain data from multiple different RPC requests for Auto-Batching.
- OP can be started by a thread or process.

#### <b>2.2 Channel Design</b>

- Channel is the data structure for sharing data between OPs, responsible for sharing data or sharing data status information.
- Outputs from multiple OPs can be stored in the same Channel, and data from the same Channel can be used by multiple OPs.
- The following illustration shows the design of Channel in the graph execution engine, using input buffer and output buffer to align data between multiple OP inputs and multiple OP outputs, with a queue in the middle to buffer.

<center>
<img src='pipeline_serving-image3.png' height = "500" align="middle"/>
</center>


#### <b>2.3 client type design</b>

- Prediction type (client_type) of Op has 3 types, brpc, grpc and local_predictor
- brpc: Using bRPC Client to interact with remote Serving by network, performance is better than grpc.
  - grpc: Using gRPC Client to interact with remote Serving by network, cross-platform deployment supported.
  - local_predictor: Load the model and predict in the local service without interacting with the network. Support multi-card deployment, and TensorRT prediction.
  - Selection: 
    - Time cost(lower is better): local_predict < brpc <= grpc
    - Microservice: Split the brpc or grpc model into independent services, simplify development and deployment complexity, and improve resource utilization

#### <b>2.4 Extreme Case Consideration</b>

- Request timeout

  The entire graph execution engine may time out at every step. The graph execution engine controls the time out by setting `timeout` value. Requests that time out at any step will return a timeout response.

- Channel stores too much data

  Channels may store too much data, causing copy time to be too high. Graph execution engines can store OP calculation results in external memory, such as high-speed memory KV systems.

- Whether input buffers and output buffers in Channel will increase indefinitely

  - It will not increase indefinitely. The input to the entire graph execution engine is placed inside a Channel's internal queue, directly acting as a traffic control buffer queue for the entire service.
  - For input buffer, adjust the number of concurrencies of OP1 and OP2 according to the amount of computation, so that the number of input buffers from each input OP is relatively balanced. (The length of the input buffer depends on the speed at which each item in the internal queue is ready)
  - For output buffer, you can use a similar process as input buffer, which adjusts the concurrency of OP3 and OP4 to control the buffer length of output buffer. (The length of the output buffer depends on the speed at which downstream OPs obtain data from the output buffer)
  - The amount of data in the Channel will not exceed `worker_num` of gRPC, that is, it will not exceed the thread pool size.

## ★ Detailed Design

#### 1. General OP Definition

As the basic unit of graph execution engine, the general OP constructor is as follows:

```python
def __init__(name=None,
             input_ops=[],
             server_endpoints=[],
             fetch_list=[],
             client_config=None,
             client_type=None,
             concurrency=1,
             timeout=-1,
             retry=1,
             batch_size=1,
             auto_batching_timeout=None,
             local_service_handler=None)
```

The meaning of each parameter is as follows:

|       Parameter       |                           Meaning                            |
| :-------------------: | :----------------------------------------------------------: |
|         name          | (str) String used to identify the OP type, which must be globally unique. |
|       input_ops       |     (list) A list of all previous OPs of the current Op.     |
|   server_endpoints    | (list) List of endpoints for remote Paddle Serving Service. If this parameter is not set,it is considered as local_precditor mode, and the configuration is read from local_service_conf |
|      fetch_list       | (list) List of fetch variable names for remote Paddle Serving Service. |
|     client_config     | (str) The path of the client configuration file corresponding to the Paddle Serving Service. |
|     client_type       | （str)brpc, grpc or local_predictor. local_predictor does not start the Serving service, in-process prediction|
|      concurrency      |             (int) The number of concurrent OPs.              |
|        timeout        | (int) The timeout time of the process operation, in ms. If the value is less than zero, no timeout is considered. |
|         retry         | (int) Timeout number of retries. When the value is 1, no retries are made. |
|      batch_size       | (int) The expected batch_size of Auto-Batching, since building batches may time out, the actual batch_size may be less than the set value. |
| auto_batching_timeout | (float) Timeout for building batches of Auto-Batching (the unit is ms). When batch_size> 1, auto_batching_timeout should be set, otherwise the waiting will be blocked when the number of requests is insufficient for batch_size|
| local_service_handler | (object) local predictor handler，assigned by Op init() input parameters or created in Op init()|


#### 2. General OP Secondary Development Interface

|              Interface or Variable               |                           Explain                            |
| :----------------------------------------------: | :----------------------------------------------------------: |
|        def preprocess(self, input_dicts)         | Process the data obtained from the channel, and the processed data will be used as the input of the **process** function. (This function handles a **sample**) |
| def process(self, feed_dict_list, typical_logid) | The RPC prediction process is based on the Paddle Serving Client, and the processed data will be used as the input of the **postprocess** function. (This function handles a **batch**) |
|  def postprocess(self, input_dicts, fetch_dict)  | After processing the prediction results, the processed data will be put into the subsequent Channel to be obtained by the subsequent OP. (This function handles a **sample**) |
|                def init_op(self)                 |      Used to load resources (such as word dictionary).       |
|               self.concurrency_idx               | Concurrency index of current process(not thread) (different kinds of OP are calculated separately). |

In a running cycle, OP will execute three operations: preprocess, process, and postprocess (when the `server_endpoints` parameter is not set, the process operation is not executed). Users can rewrite these three functions. The default implementation is as follows:

```python
def preprocess(self, input_dicts):
  # multiple previous Op
  if len(input_dicts) != 1:
    raise NotImplementedError(
      'this Op has multiple previous inputs. Please override this func.'
    ）
  (_, input_dict), = input_dicts.items()
  return input_dict

def process(self, feed_dict_list, typical_logid):
  err, err_info = ChannelData.check_batch_npdata(feed_dict_list)
  if err != 0:
    raise NotImplementedError(
      "{} Please override preprocess func.".format(err_info))
  call_result = self.client.predict(
    feed=feed_dict_list, fetch=self._fetch_names, log_id=typical_logid)
  if isinstance(self.client, MultiLangClient):
    if call_result is None or call_result["serving_status_code"] != 0:
      return None
    call_result.pop("serving_status_code")
  return call_result

def postprocess(self, input_dicts, fetch_dict):
  return fetch_dict
```

The parameter of **preprocess** is the data `input_dicts` in the previous Channel. This variable (as a **sample**) is a dictionary with the name of the previous OP as key and the output of the corresponding OP as value.

The parameter of **process** is the input variable `fetch_dict_list` (a list of the return value of the preprocess function) of the Paddle Serving Client prediction interface. This variable (as a **batch**) is a list of dictionaries with feed_name as the key and the data in the ndarray format as the value. `typical_logid` is used as the logid that penetrates to PaddleServingService.

The parameters of **postprocess** are `input_dicts` and `fetch_dict`. `input_dicts` is consistent with the parameter of preprocess, and `fetch_dict` (as a **sample**) is a sample of the return batch of the process function (if process is not executed, this value is the return value of preprocess).

Users can also rewrite the **init_op** function to load some custom resources (such as word dictionary). The default implementation is as follows:

```python
def init_op(self):
  pass
```

It should be **noted** that in the threaded version of OP, each OP will only call this function once, so the loaded resources must be thread safe.

#### 3. RequestOp Definition and Secondary Development Interface

RequestOp is used to process RPC data received by Pipeline Server, and the processed data will be added to the graph execution engine. Its constructor is as follows:

```python
def __init__(self)
```

When the default RequestOp cannot meet the parameter parsing requirements, you can customize the request parameter parsing method by rewriting the following two interfaces.

|           Interface or Variable           |                           Explain                            |
| :---------------------------------------: | :----------------------------------------------------------: |
|             def init_op(self)             | It is used to load resources (such as dictionaries), and is consistent with general OP. |
| def unpack_request_package(self, request) |                  Process received RPC data.                  |

The default implementation of **unpack_request_package** is to make the key and value in RPC request into a dictionary:

```python
def unpack_request_package(self, request):
  dictdata = {}
  for idx, key in enumerate(request.key):
    data = request.value[idx]
    try:
      data = eval(data)
    except Exception as e:
      pass
    dictdata[key] = data
  return dictdata
```

The return value is required to be a dictionary type.

#### 4. ResponseOp Definition and Secondary Development Interface

ResponseOp is used to process the prediction results of the graph execution engine. The processed data will be used as the RPC return value of Pipeline Server. Its constructor is as follows:

```python
def __init__(self, input_ops)
```

`input_ops` is the last OP of graph execution engine. Users can construct different DAGs by setting different `input_ops` without modifying the topology of OPs.

When the default ResponseOp cannot meet the requirements of the result return format, you can customize the return package packaging method by rewriting the following two interfaces.

|            Interface or Variable             |                           Explain                            |
| :------------------------------------------: | :----------------------------------------------------------: |
|              def init_op(self)               | It is used to load resources (such as dictionaries), and is consistent with general OP. |
| def pack_response_package(self, channeldata) | Process the prediction results of the graph execution engine as the return of RPC. |

The default implementation of **pack_response_package** is to convert the dictionary of prediction results into key and value in RPC response:

```python
def pack_response_package(self, channeldata):
  resp = pipeline_service_pb2.Response()
  resp.ecode = channeldata.ecode
  if resp.ecode == ChannelDataEcode.OK.value:
    if channeldata.datatype == ChannelDataType.CHANNEL_NPDATA.value:
      feed = channeldata.parse()
      np.set_printoptions(threshold=np.nan)
      for name, var in feed.items():
        resp.value.append(var.__repr__())
        resp.key.append(name)
    elif channeldata.datatype == ChannelDataType.DICT.value:
      feed = channeldata.parse()
      for name, var in feed.items():
        if not isinstance(var, str):
          resp.ecode = ChannelDataEcode.TYPE_ERROR.value
          resp.error_info = self._log(
            "fetch var type must be str({}).".format(type(var)))
          break
        resp.value.append(var)
        resp.key.append(name)
    else:
      resp.ecode = ChannelDataEcode.TYPE_ERROR.value
      resp.error_info = self._log(
        "Error type({}) in datatype.".format(channeldata.datatype))
  else:
    resp.error_info = channeldata.error_info
  return resp
```

#### 5. PipelineServer Definition

The definition of PipelineServer is relatively simple, as follows:

```python
server = PipelineServer()
server.set_response_op(response_op)
server.prepare_server(config_yml_path)
server.run_server()
```

Where `response_op` is the responseop mentioned above, PipelineServer will initialize Channels according to the topology relationship of each OP and build the calculation graph. `config_yml_path` is the configuration file of PipelineServer. The example file is as follows:

```yaml
# gRPC port
rpc_port: 18080  

# http port, do not start HTTP service when the value is less or equals 0. The default value is 0.
http_port: 18071 

# gRPC thread pool size (the number of processes in the process version servicer). The default is 1
worker_num: 1  

 # Whether to use process server or not. The default is false
build_dag_each_worker: false 

dag:
    # Whether to use the thread version of OP. The default is true
    is_thread_op: true  

    # The number of times DAG executor retries after failure. The default value is 1, that is, no retrying
    retry: 1 

    # Whether to print the log on the server side. The default is false
    use_profile: false  

    # Monitoring time interval of Tracer (in seconds). Do not start monitoring when the value is less than 1. The default value is -1
    tracer:
        interval_s: 600 

op:
    bow:
        # Concurrency, when is_thread_op=True, it's thread concurrency; otherwise, it's process concurrency
        concurrency: 1

        # Client types, brpc, grpc and local_predictor
        client_type: brpc

        # Retry times, no retry by default
        retry: 1

        # Prediction timeout, ms
        timeout: 3000

        # Serving IPs
        server_endpoints: ["127.0.0.1:9393"]

        # Client config of bow model
        client_config: "imdb_bow_client_conf/serving_client_conf.prototxt"

        # Fetch list
        fetch_list: ["prediction"]    
        
        # Batch size, default 1
        batch_size: 1

        # Batch query timeout
        auto_batching_timeout: 2000
```

### 6. Special usages

#### 6.1 <b>Business custom error type</b>

Users can customize the error code according to the business, inherit ProductErrCode, and return it in the return list in Op's preprocess or postprocess. The next stage of processing will skip the post OP processing based on the custom error code.

```python
class ProductErrCode(enum.Enum):
    """
    ProductErrCode is a base class for recording business error code. 
    product developers inherit this class and extend more error codes. 
    """
    pass
```

#### <b>6.2 Skip process stage</b>

The 2rd result of the result list returned by preprocess is `is_skip_process=True`, indicating whether to skip the process stage of the current OP and directly enter the postprocess processing

```python
def preprocess(self, input_dicts, data_id, log_id):
        """
        In preprocess stage, assembling data for process stage. users can 
        override this function for model feed features.
        Args:
            input_dicts: input data to be preprocessed
            data_id: inner unique id
            log_id: global unique id for RTT
        Return:
            input_dict: data for process stage
            is_skip_process: skip process stage or not, False default
            prod_errcode: None default, otherwise, product errores occured.
                          It is handled in the same way as exception. 
            prod_errinfo: "" default
        """
        # multiple previous Op
        if len(input_dicts) != 1:
            _LOGGER.critical(
                self._log(
                    "Failed to run preprocess: this Op has multiple previous "
                    "inputs. Please override this func."))
            os._exit(-1)
        (_, input_dict), = input_dicts.items()
        return input_dict, False, None, ""

```

#### <b>6.3 Custom proto Request and Response</b>

When the default proto structure does not meet the business requirements, at the same time, the Request and Response message structures of the proto in the following two files remain the same.

> pipeline/gateway/proto/gateway.proto 

> pipeline/proto/pipeline_service.proto

Recompile Serving Server again.

#### <b>6.4 Custom URL</b>

The grpc gateway processes post requests. The default `method` is `prediction`, for example: 127.0.0.1:8080/ocr/prediction. Users can customize the name and method, and can seamlessly switch services with existing URLs.

```proto
service PipelineService {
  rpc inference(Request) returns (Response) {
    option (google.api.http) = {
      post : "/{name=*}/{method=*}"
      body : "*"
    };
  }
};
```

***

## ★ Classic examples

Here, we build a simple imdb model enable example to show how to use Pipeline Serving. The relevant code can be found in the `python/examples/pipeline/imdb_model_ensemble` folder. The Server-side structure in the example is shown in the following figure:



<center>
<img src='pipeline_serving-image4.png' height = "200" align="middle"/>
</center>


### 1. Get the model file and start the Paddle Serving Service

```shell
cd python/examples/pipeline/imdb_model_ensemble
sh get_data.sh
python -m paddle_serving_server.serve --model imdb_cnn_model --port 9292 &> cnn.log &
python -m paddle_serving_server.serve --model imdb_bow_model --port 9393 &> bow.log &
```

PipelineServing also supports local automatic startup of PaddleServingService. Please refer to the example `python/examples/pipeline/ocr`.

### 2. Create config.yaml

Because there is a lot of configuration information in config.yaml,, only part of the OP configuration is shown here. For full information, please refer to `python/examples/pipeline/imdb_model_ensemble/config.yaml`

```yaml
op:
    bow:
        # Concurrency, when is_thread_op=True, it's thread concurrency; otherwise, it's process concurrency
        concurrency: 1

        # Client types, brpc, grpc and local_predictor
        client_type: brpc

        # Retry times, no retry by default
        retry: 1

        # Predcition timeout, ms
        timeout: 3000

        # Serving IPs
        server_endpoints: ["127.0.0.1:9393"]

        # Client config of bow model
        client_config: "imdb_bow_client_conf/serving_client_conf.prototxt"

        # Fetch list
        fetch_list: ["prediction"]    
        
        # Batch request size, default 1
        batch_size: 1

        # Batch query timeout
        auto_batching_timeout: 2000
    cnn:
        # Concurrency
        concurrency: 1

        # Client types, brpc, grpc and local_predictor
        client_type: brpc

        # Retry times, no retry by default
        retry: 1

        # Predcition timeout, ms
        timeout: 3000

        # Serving IPs
        server_endpoints: ["127.0.0.1:9292"]

        # Client config of cnn model
        client_config: "imdb_cnn_client_conf/serving_client_conf.prototxt"

        # Fetch list
        fetch_list: ["prediction"]
        
        # Batch request size, default 1
        batch_size: 1

        # Batch query timeout
        auto_batching_timeout: 2000
    combine:
        # Concurrency
        concurrency: 1

        #R etry times, no retry by default
        retry: 1

        # Predcition timeout, ms
        timeout: 3000

        # Batch request size, default 1
        batch_size: 1

        # Batch query timeout, ms
        auto_batching_timeout: 2000

### 3. Start PipelineServer

Run the following code

```python
from paddle_serving_server.pipeline import Op, RequestOp, ResponseOp
from paddle_serving_server.pipeline import PipelineServer
from paddle_serving_server.pipeline.proto import pipeline_service_pb2
from paddle_serving_server.pipeline.channel import ChannelDataEcode
import numpy as np
from paddle_serving_app.reader import IMDBDataset

class ImdbRequestOp(RequestOp):
    def init_op(self):
        self.imdb_dataset = IMDBDataset()
        self.imdb_dataset.load_resource('imdb.vocab')

    def unpack_request_package(self, request):
        dictdata = {}
        for idx, key in enumerate(request.key):
            if key != "words":
                continue
            words = request.value[idx]
            word_ids, _ = self.imdb_dataset.get_words_and_label(words)
            dictdata[key] = np.array(word_ids)
        return dictdata


class CombineOp(Op):
    def preprocess(self, input_data):
        combined_prediction = 0
        for op_name, data in input_data.items():
            _LOGGER.info("{}: {}".format(op_name, data["prediction"]))
            combined_prediction += data["prediction"]
        data = {"prediction": combined_prediction / 2}
        return data


read_op = ImdbRequestOp()
bow_op = Op(name="bow",
            input_ops=[read_op],
            server_endpoints=["127.0.0.1:9393"],
            fetch_list=["prediction"],
            client_config="imdb_bow_client_conf/serving_client_conf.prototxt",
            concurrency=1,
            timeout=-1,
            retry=1)
cnn_op = Op(name="cnn",
            input_ops=[read_op],
            server_endpoints=["127.0.0.1:9292"],
            fetch_list=["prediction"],
            client_config="imdb_cnn_client_conf/serving_client_conf.prototxt",
            concurrency=1,
            timeout=-1,
            retry=1)
combine_op = CombineOp(
    name="combine",
    input_ops=[bow_op, cnn_op],
    concurrency=5,
    timeout=-1,
    retry=1)

# use default ResponseOp implementation
response_op = ResponseOp(input_ops=[combine_op])

server = PipelineServer()
server.set_response_op(response_op)
server.prepare_server('config.yml')
server.run_server()
```

### 4. Perform prediction through PipelineClient

```python
from paddle_serving_client.pipeline import PipelineClient
import numpy as np

client = PipelineClient()
client.connect(['127.0.0.1:18080'])

words = 'i am very sad | 0'

futures = []
for i in range(3):
    futures.append(
        client.predict(
            feed_dict={"words": words},
            fetch=["prediction"],
            asyn=True))

for f in futures:
    res = f.result()
    if res["ecode"] != 0:
        print(res)
        exit(1)
```

***

## ★ Performance analysis


### 1. How to optimize with the timeline tool

In order to better optimize the performance, PipelineServing provides a timeline tool to monitor the time of each stage of the whole service.

### 2. Output profile information on server side

The server is controlled by the `use_profile` field in yaml:

```yaml
dag:
    use_profile: true
```

After the function is enabled, the server will print the corresponding log information to the standard output in the process of prediction. In order to show the time consumption of each stage more intuitively, Analyst module is provided for further analysis and processing of log files.

The output of the server is first saved to a file. Taking `profile.txt` as an example, the script converts the time monitoring information in the log into JSON format and saves it to the `trace` file. The `trace` file can be visualized through the tracing function of Chrome browser.

```shell
from paddle_serving_server.pipeline import Analyst
import json
import sys

if __name__ == "__main__":
    log_filename = "profile.txt"
    trace_filename = "trace"
    analyst = Analyst(log_filename)
    analyst.save_trace(trace_filename)
```

Specific operation: open Chrome browser, input in the address bar `chrome://tracing/` , jump to the tracing page, click the load button, open the saved `trace` file, and then visualize the time information of each stage of the prediction service.

### 3. Output profile information on client side

The profile function can be enabled by setting `profile=True` in the `predict` interface on the client side.

After the function is enabled, the client will print the log information corresponding to the prediction to the standard output during the prediction process, and the subsequent analysis and processing are the same as that of the server.

### 4. Analytical methods
```
cost of one single OP：
op_cost = process(pre + mid + post) 

OP Concurrency: 
op_concurrency = op_cost(s) * qps_expected

Service throughput：
service_throughput = 1 / slowest_op_cost * op_concurrency

Service average cost：
service_avg_cost = ∑op_concurrency in critical Path

Channel accumulations：
channel_acc_size = QPS(down - up) * time

Average cost of batch predictor：
avg_batch_cost = (N * pre + mid + post) / N 
```