But we
have already experienced failures, and some have been intermittent and difficult to
reproduce. Therefore, we have resolved to test and analyze the system thoroughly,
to discover the different ways it can fail, in order to prepare for correcting defects
or designing a more robust system.
Before we can describe our testing and analysis, we must explain a bit more about
how our programs work.
2.2 Protocol
Our two programs implement a protocol, an agreement about how to work together.
A protocol is defined by rules for forming messages and rules that constrain the
ordering of the messages. In this example, the ???messages??? in the protocol are the
Overview 15
method calls coded by developers, not the actual messages that travel over the
network. In this protocol, the server starts first, and waits for a connection from a
client. The client connects to the server and sends a command: the string T to request
a temperature measurement. The server responds by sending back the temperature,
expressed in a string such as 72.1. Then the client may send another command, or
it may close the connection and exit. If the client closes, the server may wait for
another client to connect, or it may also exit. The server can only accommodate one
client connection at a time.
2.3 Sockets
We implement our protocol using sockets, an application programming interface
(API) for communicating over TCP/IP networks.
Pages:
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46