Client-server communication#

Terminology#

DPF is based on a client-server architecture.

A DPF Server is a set of files that enables DPF capabilities.

PyDPF-Core is a Python client API communicating with a DPF Server, either directly in the same process or through the network using gRPC.

DPF Server in the same process#

Default use of a PyDPF-Core client and a DPF Server is in the same process, using the InProcess class.

from ansys.dpf import core as dpf
local_server = dpf.start_local_server()
local_server

<ansys.dpf.core.server_types.InProcessServer object at ...>

This DPF Server can now be used to instantiate models, operators, and more.

# instantiate an operator
local_operator = dpf.operators.results.displacement(server=local_server)

# instantiate a model
from ansys.dpf.core import examples
local_model = dpf.Model(examples.find_simple_bar(), server=local_server)

DPF Server through the network using gRPC#

The GrpcServer class is used to enable gRPC communication:

from ansys.dpf import core as dpf
grpc_server_config = dpf.AvailableServerConfigs.GrpcServer
grpc_server = dpf.start_local_server(config=grpc_server_config)
grpc_server

<ansys.dpf.core.server_types.GrpcServer object at ...>

You can obtain the server port and IP address:

print(grpc_server)

DPF Server: {'server_ip': '127.0.0.1', 'server_port': 50052, 'server_process_id': 9999, 'server_version': '6.0', 'os': 'nt'}

From another machine, you can connect remotely to this DPF Server and instantiate models, operators, and more:

from ansys.dpf import core as dpf
grpc_remote_server = dpf.connect_to_server(ip='127.0.0.1', port=50052)

# instantiate an operator
remote_operator = dpf.operators.results.displacement(server=grpc_remote_server)

# instantiate a model
from ansys.dpf.core import examples
remote_model = dpf.Model(examples.find_simple_bar(), server=grpc_remote_server)

Through the network using gRPC, a DPF sever enables distributed computation capabilities. For more information, see Examples for postprocessing on distributed processes.

DPF Server startup using a configuration#

The different DPF server types can be started using one of the AvailableServerConfigs configurations.

in_process_config = dpf.AvailableServerConfigs.InProcessServer
in_process_server = dpf.start_local_server(config=in_process_config)

grpc_config = dpf.AvailableServerConfigs.GrpcServer
grpc_server = dpf.start_local_server(config=grpc_config)

legacy_grpc_config = dpf.AvailableServerConfigs.LegacyGrpcServer
legacy_grpc_server = dpf.start_local_server(config=legacy_grpc_config)

Advanced concepts and release history#

The communication logic with a DPF server is defined when starting it using an instance of the ServerConfig class. Different predefined server configurations are available in DPF, each answering a different use case. For more information, see the AvailableServerConfigs class.

  • The GrpcServer configuration is available in server version 4.0 (Ansys 2022 R2) and later. It allows you to remotely connect to a DPF server across a network by telling the client to communicate with this server via the gRPC communication protocol. Although it can be used to communicate with a DPF server running on the same local machine, the next configuration is better for this option.

  • The InProcessServer configuration is available in server version 4.0 (Ansys 2022 R2) and later. It indicates to the client that a DPF server is installed on the local machine, enabling direct calls to the server binaries from within the client’s own Python process. This removes the need to copy and send data between the client and server, and it makes calls to the server functionalities much faster and uses less memory.

  • The LegacyGrpcServer configuration is the only one available for server versions 4.0 and earlier (Ansys 2022 R1, 2021 R2, and 2021 R1). The client communicates with a local or remote DPF server via the gRPC communication protocol.

For DPF with Ansys 2023 R1 and later, InProcessServer is the default configuration, which means that servers are launched on the local machine. To launch a DPF server on a remote machine and communicate with it using gRPC, use the GrpcServer configuration as shown in ref_server_types_example.