Use > Database > Introducing the Database Environment > Introduction to Preparing the Database Environment > Large Capacity Planning for UCMDB

Large Capacity Planning for UCMDB

This section includes:

ClosedLarge Capacity Planning Overview

Using the default configuration, Universal CMDB can work with a deployment of more than 25 million CIs and relationships. To work with a larger deployment, you must implement the following configuration:

Depending in the number of CIs and relationships, increase the CMDB heap as follows:

# CIs and Relationships Heap Size
≤ 40 million 12 GB
40 million – 60 million 16 GB
60 million – 125 million 24 GB
> 125 million 55 GB

  • For capacity planning requirements, see the Support Matrix section of the UCMDB Help.
  • For details about the changes you must make to the system configuration to support this capacity, see "Configuring the UCMDB Server for Large Capacity" in the Deployment Guide.
  • For details on how to improve performance, see "Configuring the Oracle Database for Large Capacity" and "Configuring the Microsoft SQL Database for Large Capacity" in the Deployment Guide.
  • For details about the setup used for capacity testing and performance results, see "System Capacity Test" in the Deployment Guide.
ClosedConfiguring the UCMDB Server for Large Capacity

For the system to support the desired number of CIs and relationships, update the following parameters on the UCMDB Server:

  Parameter Default CIs and Relationships (million)   Location
≤ 40
 40 – 60
60 – 125 > 125
wrapper.java.
initmemory 		1024 2048 8192
  • Windows: C:\UCMDB\UCMDBServer\bin\wrapper-platform.conf
  • Linux: /opt/UCMDB/UCMDBServer/bin/wrapper-platform.conf
wrapper.java.
maxmemory 		4096 8192 16384 24576 56320
wrapper.java.
additional.31=
-XX:MaxMetas
paceSize 		256 512 1024
dal.object.condi
tion.max.result.
size 		2000000 50000000 50000000
  • Windows: C:\UCMDB\UCMDBServer\conf\
    settings.override.properties
  • Linux: /opt/UCMDB/UCMDBServer/conf/
    settings.override.properties
dal.use.memory.
instead.temp.tab
le.high.threshold.
oracle 		600000 6000000 6000000 10000000
dal.joinf.max.res
ult.size 		400000 4000000 4000000
ClosedConfiguring the Oracle Database for Large Capacity

When working on a system containing more than 40 million objects and relationships, you can improve performance by increasing the Oracle SGA and PGA to the following suggested sizes:

CIs and Relationships SGA PGA
40 million – 60 million 22 GB 6 GB
60 million – 120 million 42 GB 14 GB
> 120 million 88 GB 24 GB

This improves the performance of both the TQL calculation for several types of TQL queries, as well as for data-in operations performed on the system.

ClosedConfiguring the Microsoft SQL Database for Large Capacity

When working on a system containing more than 40 million objects and relationships, you can improve performance by increasing the Microsoft SQL Server Memory to the following suggested sizes:

CIs and Relationships Microsoft SQL Server Memory
40 million – 60 million 28 GB
> 60 million 56 GB
ClosedSystem Capacity Test

Setup

The system capacity test is conducted for Microsoft SQL Server (with 125 million CIs and relationships) and Oracle Database (with 200 million CIs and relationships) separately, by using the following hardware configurations.

 

The following business flows were tested as part of the system test:

  • TQL Calculation

    TQLs were divided into sub groups according to the result size (<100, <1000, and <10000), according to the data set that the TQL retrieves, and according to the TQL configuration:

    • Like Condition
    • Like Ignore case
    • Different number of hierarchies in the TQL results (2-5)
    • Compound
    • Sub-graph

  • Data-in

    The data-in scenario in the system test included insertion, updates, and deletion.

  • Enrichments

    Enrichment scenarios included insert, update, and delete.

Results

Following the load test in the scenario that includes query execution (2 days), data-in (7 days for Oracle and 10 days for MS SQL), and enrichment execution, the following results were achieved:

  • The system was stable throughout the run. No restarts, memory leaks, or any other degradation over time was observed.
  • System performance was acceptable.

Send Help Center feedback