7.21. Network load of the IDERI note Client

If you are planning to use IDERI note in your network, you are probably interested in the network load that the IDERI note client will cause. There cannot be a simple answer to this question, because as with all networked computer applications, network load heavily depends on usage patterns. With the IDERI note client, network load also heavily depends on the client polling interval that you choose. The following three tests might give you a rough idea what the load of your network will be per client connection. All three tests have been made for a duration of roughly 8 hours and the network load was measured for the IDERI note server using NTOP (http://www.ntop.org). The client polling interval for all tests has been artificially set to 20s, so the observation period of these 8 hour tests resembles a 24 hour test with the shortest polling interval of the IDERI note client (1 minute) that can be specified with the IDERI note client installer. Each test has been conducted with 5 different clients connected to an IDERI note server using a named pipes connection.

The first test measures the network traffic per each client if no new messages are created during the observation period. That means that clients do not download new messages and just periodically check if new messages are available for them. Figure 7.80 shows the network traffic that has been measured throughout the observation period for the TCP sessions that have been established to the IDERI note server.

Figure 7.80: Network traffic per each client to the IDERI note server if no new messages are created

As you can see, the client sends around 400 kBytes of data and receives around 300 kBytes of data. With our extrapolation to 24 hours, this makes 285 of bytes sent and 214 of bytes received per minute or per polling interval.

The second test is similar to the first test, but this time it measures network traffic per each client if a new message of size 80 characters is created every 10 seconds. This would resemble a rate of 2 new messages per minute if we extrapolate the 8 hour test to a 24 hour test as described above. This means that during each polling interval, the client receives two such messages.

Figure 7.81 shows the network traffic that has been measured throughout the observation period in this test for the TCP sessions that have been established to the IDERI note server.

Figure 7.81: Network traffic per each client to the IDERI note server if a new message is created every 10 seconds

In this test, each client sends 1.3 MBytes and receives 1.6 MBytes. In order to account for rounding errors, this is 1.4*1024*1024=1468007 Bytes sent and 1.7*1024*1024=1782580 Bytes received. With our extrapolation to 24 hours, this is 61167 Bytes sent and 74275 Bytes received per hour or 1020 Bytes sent and 1238 Bytes received per minute or per polling interval.

The third test is quite similar to the second test. Again, it measures network traffic per each client if a new message of size 80 characters is created every 10 seconds, but this time, clients have to notify the server that they have received the message.

Figure 7.82 shows the network traffic that has been measured throughout the observation period in this test for the TCP sessions that have been established to the IDERI note server.

Figure 7.82: Network traffic per each client to the IDERI note server if a new message is created every 10 seconds and clients have to notify the server that they received the message

In this test, each client sends and receives around 2.2 MBytes. In order to account for rounding errors, this is 2.3*1024*1024=2411725 Bytes. With our extrapolation to 24 hours, this is 100489 Bytes sent and received per hour or 1675 Bytes sent and received per minute or per polling interval.