UMP/USM performance - Checklist of factors that need to be considered that affect 'Loading...' and/or display of metrics within a reasonable time frame
Article ID: 127497
DX Infrastructure ManagementNIMSOFT PROBES
The following knowledge base article provides a summarized checklist of factors that need to be considered when analyzing UMP performance/response time - Factors that affect 'Loading...' and/or display of metrics within a reasonable time frame. This information can help customers and support collect and analyze more complete data to be used in optimizing UMP/USM performance. Customers should work closely with their DBA regarding most of the factors listed below.
How can we analyze/minimize UMP/USM response time?
- UIM 8.5.1 or higher
- Identify where in the UMP interface/USM portlet is it 'slow', e.g., Loading... phase? or for specific views, specific probe metrics? PRD/List reports? How many seconds? - Consistent OR intermittent slowness? - UIM/UMP/DB hardware requirements followed? - Type of storage, Tier 1? - SSD drives? - RAID 10 configuration? - Database type (Enterprise versus Standard), Enterprise is recommended. - UIM Service Pack applied? - UMP/USM hotfixes applied? - Database Partitioning enabled (database and data_engine) - Enterprise ONLY. - Fragmentation of key tables (CM_*, S_QOS_*, NAS_*) - daily job should be run to defrag specific tables. - Any SQL Server memory pressure on the database over time? - What is the total database size? - How many large tables? Large tables (> 100 million rows) - data_engine data retention settings being adhered to? - USM queries: USM debug can be enabled to uncover any/all slow queries which may require some additional indexes to be created - Check resource utilization over time on the database server and UMP machine, CPU, Memory , Disk I/O, network speed/available bandwidth, proxy?, memory dedicated to SQL Server?, AV blocking/scanning? Database on the same subnet as the Primary hub? - wasp Java heap memory utilization versus what is configured - Monitoring governance considerations: too much non-critical/unnecessary data being collected? monitoring/polling intervals too low (frequent) for some QOS? - This can be overridden for specific QOS in the data_engine. - Test using different browsers after deleting the cache