Effect of the network on performance

Edit online

The network that connects the hub server and the system where the Operations Center is installed can affect the performance of the Operations Center.

To achieve better system performance, use one of the following configurations:
  • The Operations Center is on the same system as the hub server.
  • The Operations Center is on a system that is located physically near the hub server system.

Also, consider facilitating upgrades and maintenance by designating a physical or virtual system that is separate from the production environment as the hub server.

Network latency

Network latency is the time interval between the following operations:
  • The initiation of a send operation from a source system
  • The completion of the matching receive operation by the target system
Latency between the Operations Center web server and web browsers
For the best responsiveness when logged in to the Operations Center, use a network connection with a round-trip latency that is no greater than 5 ms. This latency can typically be achieved when the systems are on the same local area network (LAN). Higher latencies might be acceptable but can cause degraded responsiveness. For example, the responsiveness across a wide area network (WAN) might not be acceptable to users of the Operations Center.
Latency between the Operations Center web server and the hub server
For the best responsiveness, use a network connection with a round-trip latency that is no greater than 10 ms. Lowest latency is achieved when both of these servers are installed on the same system or on the same LAN.
Latency between the hub server and spoke servers
Round-trip latency can be high, for example, 100 ms, or low, for example, 10 ms. However, with high latency, Operations Center pages that show details about an individual client, policy domain, server, or storage pool might take a longer time to be displayed. Also, if connection timeout issues occur, you might be required to adjust the ADMINCOMMTIMEOUT value on the hub and spoke servers. For information about what to do if connection timeout issues occur, see the following known issue on the IBM® Support Portal at technote 497591.

Estimating the network latency

You can use a ping command to estimate the network latency. To use a ping command, complete the following steps:
  1. From the source system, ping the target system by issuing the following commands, where remoteHost represents the address for the target system:
    • AIX operating systemsLinux operating systemsping –c 20 remoteHost
    • Windows operating systemsping –n 20 remoteHost
  2. Calculate the average of the intervals for all successful responses from the remote host. This calculation is an estimate of the round-trip latency.