AMPS uses the name of the client as a session identifier and as part of the identifier for messages originating from that client.

For this reason, when a transaction log is enabled in the AMPS instance (that is, when the instance is recording a sequence of publishes and attempting to eliminate duplicate publishes), an AMPS instance will only allow one application with a given client name to connect to the instance.

When a transaction log is present, AMPS requires the client name for a publisher to be:

• Unique within a set of replicated AMPS instances

• Consistent from invocation to invocation if the publisher will be publishing the same logical stream of messages

If publishers do not meet this contract (for example, if the publisher changes its name and publishes the same messages, or if a different publisher uses the same session name), message loss or duplication can happen.

60East recommends always using consistent, unique client names. For example, the client name could be formed by combining the application name, an identifier for the host system, and the ID of the user running the application. A strategy like this provides a name that will be different for different users or on different systems, but consistent for instances of the application that should be treated as equivalent to the AMPS system.

Likewise, if a publisher is sending a completely independent stream of messages (for example, a microservice that sends a different, unrelated sequence of messages each time it connects to AMPS), there is no need for a publisher to retain the same name each time it starts. However, if a publisher is resuming a stream of messages (as in the case when using a file-backed publish store), that publisher must use the same client name, since the publisher is resuming the session.

When specifying a URI for connection to an AMPS server, you may specify a number of transport-specific options in the parameters section of the URI connection parameters. Here is an example:

tcp://localhost:9007/amps/json?tcp_nodelay=true&tcp_sndbuf=100000

In this example, we have specified the AMPS instance on localhost, port 9007, connecting to a transport that uses the amps protocol and sending JSON messages. We have also set two parameters: tcp_nodelay, a Boolean (true/false) parameter, and tcp_sndbuf, an integer parameter. Multiple parameters may be combined to finely tune settings available on the transport. Normally, you'll want to stick with the defaults on your platform, but there may be some cases where experimentation and fine-tuning will yield higher or more efficient performance.

The AMPS client supports the value of tcp in the scheme component connection string for TCP/IP connections, and the value of tcps as the scheme for SSL encrypted connections.

For connections that use Unix domain sockets, the client supports the value of unix in the scheme, and requires an additional option, as described in the Unix Transports Parameters section below.

IPv6 Connections

Starting with version 5.3.3.0, the AMPS client supports creating connections over both IPv4 and IPv6 protocols if supported by the underlying Operating System.

By default, the AMPS client will prefer to resolve host names to IPv4 addresses, but this behavior can be adjusted by supplying the ip_protocol_prefer transport option, described in the table below.

TCP and SSL Transport Options

The following transport options are available for TCP connections:

Unix Transport Parameters

The unix transport type communicates over Unix domain sockets. This transport requires the following additional option:

Unix domain sockets always connect to the local system. When the scheme specified is unix, the host address is ignored in the connection string. For example, the connection string:

unix://localhost:0/amps/json?path=/sockets/the-amps-socket

and the connection string:

unix://unix:unix/amps/json?path=/sockets/the-amps-socket

are equivalent.

The other components of the connection string, including the protocol, message type, username, and authentication token are processed just as they would be for TCP/IP sockets.

AMPS Additional Logon Options

The connection string can also be used to pass logon parameters to AMPS. AMPS supports the following additional logon option:

When a client logs on to AMPS, the client sends AMPS a username and password. The username is derived from the URI, using the standard syntax for providing a user name in a URI, for example, tcp://JohnDoe:@server:port/amps/messagetype to include the user name JohnDoe in the request.

For a given user name, the password is provided by an Authenticator. The AMPS client distribution includes a DefaultAuthenticator that simply returns the password, if any, provided in the URI. A logon() command that does not specify an Authenticator will use an instance of DefaultAuthenticator.

If your authentication system requires a different authentication token, you can implement an Authenticator that provides the appropriate token.

Providing Credentials in a Connection String

When using the DefaultAuthenticator, the AMPS clients support the standard format for including a username and password in a URI, as shown below:

tcp://user:password@host:port/protocol/message_type

When provided in this form, the default authenticator provides the username and password specified in the URI. If you have implemented another authenticator, that authenticator controls how passwords are provided to the AMPS server.

In this section, we will learn more about the structure and features of the AMPS Python client and build our first Python program using AMPS.

About the Client Library

The AMPS client library is packaged as a single binary file. The exact name of the file depends on the Python version and build environment. You can find the file under the build directory of your AMPS Python client install once you've completed the build process. If you have used a prepackaged .egg or .whl to install the Python client, the appropriate binary file will be installed in your Python environment.

Every Python application you build will need to be able to reference the library. If you choose to install the python client into your local Python installation, then Python has access to the client library in the installation, and you do not need to include the library with each specific script. Otherwise, you will need to package and include the library with your script and ensure that the library is in the path where python looks for shared libraries, or ensure that the system where your application will run has installed the appropriate prepackaged build.

Connecting to AMPS

Let's begin by writing a simple program that connects to an AMPS server and sends a single message to a topic. This code will be covered in detail just following the example.

import AMPS
import sys

uri_ = "tcp://127.0.0.1:9007/amps/json"

client = AMPS.Client("examplePublisher")

try:
    client.connect(uri_)
    client.logon()
    client.publish("messages", '{ "hi" : "Hello, world!"}')

except AMPS.AMPSException as e:
    sys.stderr.write(str(e))

client.publish_flush()
client.close()

In the example above, we show the entire program. Future examples will isolate one or more specific portions of the code.

Examining the Code

Let us revisit the code we listed above:

# These import the AMPS and sys packages. All programs written using the AMPS
# Python client will need to include the import AMPS statement at a minimum.
import AMPS
import sys

# The URI to use to connect to AMPS. The URI consists of the transport, the
# address, and the protocol to use for the AMPS connection. In this case, the
# transport is tcp, the address is 127.0.0.1:9007, and the protocol is amps. This
# connection will be used for JSON messages. Check with the person who manages the
# AMPS instance to get the connection string to use for your programs.
uri_ = "tcp://127.0.0.1:9007/amps/json"


# This line creates a new Client object. Client encapsulates a single connection
# to an AMPS server. Methods on Client allow for connecting, disconnecting,
# publishing, and subscribing to an AMPS server. The argument to the Client
# constructor, "examplePublisher", is a name chosen by the client to
# identify itself to the server. Errors relating to this connection will be logged
# with reference to this name, and AMPS uses this name to help detect duplicate
# messages. AMPS enforces uniqueness for client names when a transaction log is
# configured, and it is good practice to always use unique client names.
client = AMPS.Client("examplePublisher")

# Here, we open a try block that concludes with except AMPS.AMPSException as e.
# All exceptions in AMPS derive from AMPSException. If an operation throws another
# exception, AMPS will wrap that exception into the AMPSException you receive. For
# example, if the call to connect() fails because the provided address is not
# reachable, the AMPSException will contain an inner exception from the operating
# system, likely a SocketException.
try:

    # With this statement, we provide the URI to the client and declare the AMPS
    # connection.
    client.connect(uri_)

    # The AMPS logon() command connects to AMPS and creates a named connection.
    #
    # This version of logon() uses the DefaultAuthenticator. If we
    # had provided logon credentials in the URI, that Authenticator would pass those
    # credentials to AMPS. Without credentials, the client logs on to AMPS
    # anonymously. AMPS versions 5.0 and later require a logon() command in the
    # default configuration.
    #
    # If you need to provide credentials in a different way, implement an Authenticator
    # and pass that Authenticator here in the 'authenticator' parameter.
    client.logon()

    # Here, we publish a single message to AMPS on the messages topic, containing the
    # data Hello, world!. This data is placed into a JSON message and sent to the
    # server. Upon successful completion of this function, the AMPS client has
    # enqueued the message to be sent to the server, and subscribers to the messages
    # topic will receive this JSON message: { "hi" : "Hello, world!" }.
    client.publish("messages", '{ "hi" : "Hello, world!"}')

except AMPS.AMPSException as e:
    sys.stderr.write(str(e))


# Here, we call publish_flush() on the client. This command waits until all messages
# from the client have been sent. If messages may still be in the process of
# transmission when your application is ready to close the client, publish_flush()
# helps ensure that the messages have been sent. In this case, we allow
# publish_flush() to block indefinitely. A production application might specify a
# timeout, to avoid hanging in the event that the application loses connectivity
# before the messages have been sent.
client.publish_flush()

# We close the connection to AMPS. While that doesn't matter here, since the
# application exits immediately after calling close(), it's good practice to close
# connections when you are done using them.
client.close()

The AMPS clients use connection strings to determine the server, port, transport, and protocol to use to connect to AMPS. When the connection point in AMPS accepts multiple message types, the connection string also specifies the precise message type to use for this connection.

Connection strings have a number of elements:

As shown in the figure above, connection strings have the following elements:

• Transport - Defines the network used to send and receive messages from AMPS. In this case, the transport is tcp. For connections to transports that use the Secure Sockets Layer (SSL), use tcps. For connections to AMPS over a Unix domain socket, use unix.

• Host Address - Defines the destination on the network where the AMPS instance receives messages. The format of the address is dependent on the transport. For tcp and tcps, the address consists of a host name and port number. In this case, the host address is localhost:9007. For unix domain sockets, a value for hostname and port must be provided to form a valid URI, but the content of the hostname and port are ignored, and the file name provided in the path parameter is used instead (by convention, many connection strings use localhost:0 to indicate that this is a local connection that does not use TCP/IP).

• Protocol - Sets the format in which AMPS receives commands from the client. Most code uses the default amps protocol, which sends header information in JSON format. AMPS supports the ability to develop custom protocols as extension modules, and AMPS also supports legacy protocols for backward compatibility.

• MessageType - Specifies the message type that this connection uses. This component of the connection string is required if the protocol accepts multiple message types and the transport is configured to accept multiple message types. If the protocol does not accept multiple message types, this component of the connection string is optional, and defaults to the message type specified in the transport.

  Legacy protocols such as fix, nvfix and xml only accept a single message type, and therefore do not require or accept a message type in the connection string.

As an example, a connection string such as:

tcp://localhost:9007/amps/json

would work for programs connecting from the local host to a Transport configured as follows:

<AMPSConfig>
    ...

    <!-- This transport accepts any known message type for the instance: the
         client must specify the message type. -->
    <Transport>
        <Name>any-tcp</Name>
        <Type>tcp</Type>
        <InetAddr>9007</InetAddr>
        <Protocol>amps</Protocol>
    </Transport>

    ...
</AMPSConfig>

See the AMPS Configuration Guide for more information on configuring transports.