Memory and disk requirements

IBM Tivoli Monitoring > Version 6.3 Fix Pack 2 > Installation Guides > Installation Guide > Hardware and software requirements > Required hardware for distributed systems
IBM Tivoli Monitoring, Version 6.3 Fix Pack 2

Memory and disk requirements

The following table shows estimated memory and disk storage for IBM Tivoli Monitoring components on distributed systems. The estimated memory usage for each component does not include the operating system and any concurrently running applications. To estimate the memory requirements for a machine, add the memory requirements for each component that will run on the machine, and add at least 1 GB for the operating system and to allow memory for filesystem caching.
Add the sizings for individual components to calculate a total for more than one component installed on the same computer. For example, if the hub monitoring server and portal server are installed on the same machine, the base component memory requirement is 1600 MB with an additional 220 MB per 1000 managed systems. For an environment with 2000 managed systems, the combined component memory requirement is 1600 + 2 * 220= 2040 MB. The machine should be configured with at least 4 GB RAM (2040 MB + 1 GB for the OS and filesystem caching + extra memory for dynamic requests and some situation evaluation on the hub monitoring server).
Estimated memory and disk storage for IBM Tivoli Monitoring components on distributed systems
Component Memory requirements ¹ Disk storage requirements²
Base memory requirement Additional memory requirements
Tivoli Enterprise Monitoring Server 100 MB 120 MB per 1000 managed systems in environment.
Extra memory for situations evaluated at monitoring server. ⁶
Extra memory for large dynamic data requests. ⁷
1.5 GB ^{3 5}
Remote Tivoli Enterprise Monitoring Server 100 MB 120 MB per 100 managed systems connecting through the server.
Extra memory for situations evaluated at monitoring server.
1.0 GB xLinux ^{3 5}
1.5 GB All other systems ^{3 5}
Tivoli Enterprise Monitoring Automation Server 100 MB 70 MB per 1000 managed systems in environment. 100 MB this does not include all of the additional Jazz for Service Management components ⁸
Tivoli Enterprise Portal Server 1500 MB 100 MB per 1000 managed systems in environment (for AIX systems, use 150 MB per 1000 managed systems). ⁴
Extra 800 MB if the dashboard data provider is enabled.
Extra memory for large dynamic data requests. ⁷
4.0 GB plus an additional 1.2 GB in your computer's temporary directory to install the eWAS server and the Eclipse Help Server ⁵
Portal client (browser or desktop) 400 MB Extra 200 MB for large environments. 150 MB
Tivoli Data Warehouse 4 GB depending on database configuration parameters 8 - 16 GB depending on database configuration parameters See Estimate the required size of your database.
Warehouse Proxy Agent 200 MB 150 MB
Summarization and Pruning Agent 500 MB 150 MB
Tivoli Performance Analyzer 200 MB Add 1.5 GB for large environments. Depends on Tivoli Data Warehouse configuration for Tivoli Performance Analyzer attribute groups
Dashboard Application Services Hub server (for use with IBM Infrastructure Management Dashboards for Servers) 1000 MB 3.0 GB ⁸
The memory and disk sizings shown in this table are the amounts required for the individual component beyond the needs of the operating system and any concurrently running applications. For the total system memory required by small, medium-size, and large environments, see Sizing the Tivoli Monitoring hardware.

The disk storage estimates apply to any size monitoring environment and are considered high estimates. The size of log files affect the amount of storage required.

The disk storage requirements for the hub and remote monitoring servers do not include storage for the agent depot, which can require an additional 1 GB or more.
The memory requirement for the portal server does not include database processes for the portal server database. DB2 memory usage varies from machine to machine, and can be controlled through various tuning parameters, including bufferpool size. DB2 systems frequently have self-tuning enabled to control the bufferpool size. On systems with a large amount of memory this can result in bufferpools that are much larger than necessary for the TEPS database. A good starting bufferpool size for the TEPS database is 10,000 4K pages (about 40MB). Manual control of the TEPS bufferpool size is recommended to reduce memory usage. You can use the DB2 ALTER command to change the buffer pool size. For example:
CONNECT TO TEPS;
ALTER BUFFERPOOL IBMDEFAULTBP IMMEDIATE SIZE 10000;
CONNECT RESET;
Tivoli Monitoring components require additional disk storage when the self-describing agent feature is enabled. The self-describing agent disk storage usage sizes should be multiplied by the number of products in your environment that are expected to perform self-describing agent installations. This estimate is only applicable when the self-describing agent feature is enabled. The disk storage is used by the directory indicated in the Tivoli Enterprise Monitoring Server variable TEMS_MANIFEST_PATH, or the Tivoli Enterprise Portal Server variable TEPS_MANIFEST_PATH.

The hub monitoring server requires an estimated 3 to 5 MB of additional disk storage for each product that is performing self-describing agent installations. WebSphere and OMEGAMON agents require more storage, so use the 5 MB per product estimate if you are installing those agents. Otherwise, you can use the 3 MB average size per product.

The remote monitoring server requires an estimated 2 MB of additional disk storage for each product that is performing self-describing agent installations.

The portal server requires an estimated 3 to 6 MB of additional disk storage for each product that is performing self-describing agent installations since it maintains the new and previous version of an agent product's application support. WebSphere and OMEGAMON agents require more storage, so use the 6 MB per product estimate if you are installing those agents. Otherwise, you can use the 3 MB average size per product.

Use the following formula to estimate the additional memory required for a situation evaluated at the monitoring server: 2 * number of agents running the situation * number of row instances evaluated per situation interval * row size for attribute group. The row size for an attribute group is available in the agent user's guide. As an example, assume an environment has 3000 Linux agents, each running a situation that uses the COUNT function to calculate the number of disks reported by the Linux_Disk attribute group. Assume also that the agents report on average of 15 rows for Linux_Disk, and the row size for Linux_Disk is 250 bytes. The additional memory required at the monitoring server for this situation would be: 2 * 3000 * 15 * 250 = 22,500,000 bytes.

Requests to the monitoring server for a large amount of data can increase the dynamic memory usage significantly. For example, if a user request results in 10,000 rows of Linux_Disk data (250 bytes each) being returned, the additional memory usage would be: 10,000 * 250 = 2,500,000 bytes. In sizing the memory requirements for the monitoring server, consider increasing the memory estimate by 50% to allow room for dynamic memory requests.

See the Jazz for Service Management hardware and software requirements at http://pic.dhe.ibm.com/infocenter/tivihelp/v3r1/topic/com.ibm.psc.doc_1.1.0/install/psc_c_install_prereqs.html.
Parent topic:
Required hardware for distributed systems

+
Search Tips | Advanced Search

Component	Memory requirements ¹	Disk storage requirements²
Base memory requirement	Additional memory requirements
Tivoli Enterprise Monitoring Server	100 MB	120 MB per 1000 managed systems in environment. Extra memory for situations evaluated at monitoring server. ⁶ Extra memory for large dynamic data requests. ⁷	1.5 GB ^{3 5}
Remote Tivoli Enterprise Monitoring Server	100 MB	120 MB per 100 managed systems connecting through the server. Extra memory for situations evaluated at monitoring server.	1.0 GB xLinux ^{3 5} 1.5 GB All other systems ^{3 5}
Tivoli Enterprise Monitoring Automation Server	100 MB	70 MB per 1000 managed systems in environment.	100 MB this does not include all of the additional Jazz for Service Management components ⁸
Tivoli Enterprise Portal Server	1500 MB	100 MB per 1000 managed systems in environment (for AIX systems, use 150 MB per 1000 managed systems). ⁴ Extra 800 MB if the dashboard data provider is enabled. Extra memory for large dynamic data requests. ⁷	4.0 GB plus an additional 1.2 GB in your computer's temporary directory to install the eWAS server and the Eclipse Help Server ⁵
Portal client (browser or desktop)	400 MB	Extra 200 MB for large environments.	150 MB
Tivoli Data Warehouse	4 GB depending on database configuration parameters	8 - 16 GB depending on database configuration parameters	See Estimate the required size of your database.
Warehouse Proxy Agent	200 MB		150 MB
Summarization and Pruning Agent	500 MB		150 MB
Tivoli Performance Analyzer	200 MB	Add 1.5 GB for large environments.	Depends on Tivoli Data Warehouse configuration for Tivoli Performance Analyzer attribute groups
Dashboard Application Services Hub server (for use with IBM Infrastructure Management Dashboards for Servers)	1000 MB		3.0 GB ⁸
The memory and disk sizings shown in this table are the amounts required for the individual component beyond the needs of the operating system and any concurrently running applications. For the total system memory required by small, medium-size, and large environments, see Sizing the Tivoli Monitoring hardware. The disk storage estimates apply to any size monitoring environment and are considered high estimates. The size of log files affect the amount of storage required. The disk storage requirements for the hub and remote monitoring servers do not include storage for the agent depot, which can require an additional 1 GB or more. The memory requirement for the portal server does not include database processes for the portal server database. DB2 memory usage varies from machine to machine, and can be controlled through various tuning parameters, including bufferpool size. DB2 systems frequently have self-tuning enabled to control the bufferpool size. On systems with a large amount of memory this can result in bufferpools that are much larger than necessary for the TEPS database. A good starting bufferpool size for the TEPS database is 10,000 4K pages (about 40MB). Manual control of the TEPS bufferpool size is recommended to reduce memory usage. You can use the DB2 ALTER command to change the buffer pool size. For example: CONNECT TO TEPS; ALTER BUFFERPOOL IBMDEFAULTBP IMMEDIATE SIZE 10000; CONNECT RESET; Tivoli Monitoring components require additional disk storage when the self-describing agent feature is enabled. The self-describing agent disk storage usage sizes should be multiplied by the number of products in your environment that are expected to perform self-describing agent installations. This estimate is only applicable when the self-describing agent feature is enabled. The disk storage is used by the directory indicated in the Tivoli Enterprise Monitoring Server variable TEMS_MANIFEST_PATH, or the Tivoli Enterprise Portal Server variable TEPS_MANIFEST_PATH. The hub monitoring server requires an estimated 3 to 5 MB of additional disk storage for each product that is performing self-describing agent installations. WebSphere and OMEGAMON agents require more storage, so use the 5 MB per product estimate if you are installing those agents. Otherwise, you can use the 3 MB average size per product. The remote monitoring server requires an estimated 2 MB of additional disk storage for each product that is performing self-describing agent installations. The portal server requires an estimated 3 to 6 MB of additional disk storage for each product that is performing self-describing agent installations since it maintains the new and previous version of an agent product's application support. WebSphere and OMEGAMON agents require more storage, so use the 6 MB per product estimate if you are installing those agents. Otherwise, you can use the 3 MB average size per product. Use the following formula to estimate the additional memory required for a situation evaluated at the monitoring server: 2 * number of agents running the situation * number of row instances evaluated per situation interval * row size for attribute group. The row size for an attribute group is available in the agent user's guide. As an example, assume an environment has 3000 Linux agents, each running a situation that uses the COUNT function to calculate the number of disks reported by the Linux_Disk attribute group. Assume also that the agents report on average of 15 rows for Linux_Disk, and the row size for Linux_Disk is 250 bytes. The additional memory required at the monitoring server for this situation would be: 2 * 3000 * 15 * 250 = 22,500,000 bytes. Requests to the monitoring server for a large amount of data can increase the dynamic memory usage significantly. For example, if a user request results in 10,000 rows of Linux_Disk data (250 bytes each) being returned, the additional memory usage would be: 10,000 * 250 = 2,500,000 bytes. In sizing the memory requirements for the monitoring server, consider increasing the memory estimate by 50% to allow room for dynamic memory requests. See the Jazz for Service Management hardware and software requirements at http://pic.dhe.ibm.com/infocenter/tivihelp/v3r1/topic/com.ibm.psc.doc_1.1.0/install/psc_c_install_prereqs.html.