IBM Tivoli Monitoring > Version 6.3 Fix Pack 2 > Installation Guides > Installation Guide > Pre-deployment phase > Sizing the Tivoli Monitoring hardware
IBM Tivoli Monitoring, Version 6.3 Fix Pack 2
Locating and sizing the remote Tivoli Enterprise Monitoring Server
When determining the location of your remote monitoring server, you must consider the number of connected agents, existence, number, and placement of firewalls, and network bandwidth. With the Tivoli Monitoring V6.3 release, as a general planning guideline, a single remote monitoring server can support up to 1500 managed systems.
The main factor affecting remote monitoring server scalability is situation processing, and the volume and frequency of agent attribute group data that is processed by the monitoring server during situation evaluation. Situation processing affects monitoring server performance, so the CPU and memory usage of the kdsmain process should be monitored. For tuning recommendations for improving situation processing performance, see Optimizing situations.
Although the steady state CPU utilization is usually low, a remote monitoring server uses a significant amount of system resources in processing transient workloads, such as startup, agent login, and situation distribution. Dual processor systems are recommended for remote monitoring servers. To support 1500 managed systems, a server with 4 GB of memory is typically sufficient. For environments with more than 3000 agents collecting historical data, you can use multiple Warehouse Proxy agents, with the Warehouse Proxy agents located on each remote monitoring server to handle historical data exports from agents connected to the remote monitoring server.
If the remote deployment capability is used, ensure that the remote monitoring server has sufficient network capabilities (100 Megabits per second or higher is typically sufficient). Because the remote deployment packages are large, in some cases hundreds of megabytes, you must ensure that you do not have a network bottleneck.
The network connection speed between the monitoring server and the monitoring agent will affect the elapsed time for remote deployment operations. One way to deal with slow network links is to deploy a remote monitoring server near the monitored servers. By co-locating the remote monitoring server with the agents, the Remote Deploy bundles need to be transferred across the slow link only once. Distribution to the agents can then be done on a high speed LAN segment.
Parent topic:
Sizing the Tivoli Monitoring hardware