General sizing guidelines and best practices for Infrastructure Management
Below are general guidelines for sizing the Infrastructure Management Server and the Integration Service host.
- General sizing guidelines for the Infrastructure Management Server
- General sizing guidelines for the Integration Service
Specific recommendations are given in the following topics:
Infrastructure Management Server general sizing guidelines
A single large TrueSight Infrastructure Management Server host computer scales to support the following maximum values. The values include all processes and the event management cell.
Data for 1.7 Million performance parameters
- 20,000 devices
- 250,000 monitored instances
- 3 days of raw data
- 3 months of rate data
- 40,000 intelligent events per day
- 350,000 external events per day
11.3.03 version note
From version 11.3.03, you might see 20, 30, or 50 additional monitor instances on top of the existing allowed maximum monitor instances based on the small, medium, or large environment.
These values are all independent. If you attain any of these values in a single Infrastructure Management Server instance, you must consider that you have reached the maximum capacity for that server instance. This sizing is for a 64-bit host with 8 CPUs and 32 GB of RAM. Do not try to exceed 250,000 monitor instances or 1,700,000 attributes on a single Infrastructure Management Server host. Additional CPU and memory do not help to scale beyond these values.
In some environments, the Infrastructure Management database requires more storage space than previous versions. This is because of the increased rate of performance data collection. In earlier Infrastructure Management versions, performance data was collected every 5 minutes by default. In this version of Infrastructure Management, the data collected by the BMC PATROL Agent is streamed to the TrueSight Infrastructure Management Server and many parameters are collected every minute. You must allocate 600 GB of storage (100 GB for the Infrastructure Management Server + 500 GB for the database) in a large implementation.
Integration Service general sizing guidelines
A single large Integration Service host scales to support the following maximum values. These values include the Integration Service and the event management cell.
- Data for 1.7 million performance parameters
- 900 PATROL Agents
- 250,000 monitored instances
- 25 events per second
These values are all independent. If you attain one of these values in a single Integration Service instance, consider that you have reached the maximum capacity for that Infrastructure Management Server instance.
Consider the following guidelines when sizing and deploying the Integration Service:
- A 64-bit Integration Service host machine with 2 CPUs and 2 GB RAM can scale up to 500 PATROL Agents.
Do not try to exceed 250,000 monitor instances and 1,700,000 attributes on a single Integration Service computer. Additional CPU and memory does not help to scale beyond these values.
- Allocate a separate Integration Service for each remote network.
- A 64-bit Integration Service host with 4 CPUs and 8 GB RAM can scale up to 900 PATROL Agents.
- Minimize the number of Integration Service that you use to collect data from PATROL Agents.
- Ensure that the Infrastructure Management Central and Child Servers have user access to the Integration Service computers.
- Deploy an Integration Service on each network where PATROL Agents are located, and distribute PATROL Agents on the Integration Services so that the load on the service does not exceed the suggested values.
- The load can be distributed across one or more Integration Services by keeping the total load on the server constant as per the deployment type. You can configure a maximum of 25 Integration Services.
- You can configure a maximum of 500 PATROL Agents per Integration Service in a small deployment and 900 in medium and large deployments.
Sizing and scalability best practices for all types of deployments
- Install the TrueSight Infrastructure Management Server and Integration Service on separate computers.
- Distribute event correlation, deduplication, and normalization on Infrastructure Management remote cells.
- (SAP SQL Anywhere only) To improve I/O throughput, use separate disks for the operating systems on which the TrueSight Infrastructure Management Server and the database are deployed. Ensure that each disk has its own disk controllers.
- Distribute data collection on Integration Services so that the server is dedicated to the primary server processes (analytical engine, object cache, and agent controller). The Self-Monitoring Service, which is automatically installed on the TrueSight Infrastructure Management Server, collects metrics only about the performance of the TrueSight Infrastructure Management Server.
- If you expect the total number of attributes for the environment to grow beyond 750,000 attributes, use a large-environment deployment so that the system can accommodate growth smoothly.