With NetBackup 7.5, more information is being kept in the embedded relational database. The issues that improve the performance of NetBackup master are outlined in this article.
Status Code: 252 Cannot Connect to Database
Slow Catalog backup
Sluggish or hung NetBackup job throughput
The following are the key factors for reviewing the NetBackup master system health.
Growth of EMM_DATA.db
EMM_DATA.db contains a very small amount of data but is highly transactional. Symantec NetBackup use of isolation levels requires the database to take copies of pages of the database which are marked for cleanup during system quiet times. The database will clean the pages but will not shrink disk space allocated. A large EMM_DATA.db by itself does not indicate an issue. It can be expected that this database can grow to sizes greater than 3GB for customers who have a high volume of backups even if the underlying data is only a handful of Megabytes. A database rebuild should be performed on databases higher than 3GB to reclaim the space and verify integrity. To rebuild the database, follow the instructions here:
The EMM_DATA.db growth resulting in a > 3GB is addressed in NetBackup 184.108.40.206. Customers should download and apply NetBackup 220.127.116.11 to their master server.
18.104.22.168 includes ET3123528 which addresses the following issues:
- Use isolation level 'repeatable read' instead of 'serializable' in nbpem
- Removes isolation levels that are not required from several bpdbm queries
- Releases the snapshot isolation queries before calling bpcd removing a potential delay in snapshot clean up
These fixes are required to remove the impact of a Sybase defect which will cause the snapshot isolation table to be maintained inefficiently and cause EMM_DATA.db growth.
The Sybase fix is included in NetBackup 22.214.171.124.
Additional System Configurations for improved performance:
Database Memory Allocation
NetBackup 7.5 allocates initially 1GB of RAM for the underlying database. This can be inadequate and can be adjusted to 30% of available system memory by increasing the -ch value in the server.conf. More information on the location and meaning of the server.conf values can be found in the article below:
Oracle Backup Format
Ensure that the format specified for all RMAN backup piece names, except for autobackups of the control file, ends with a _%t as documented in the NetBackup for Oracle manual. Failure to add the timestamp results in a series of extended queries that can cause significant performance degredation. These Oracle best practices and others can be found in the article below:
Semaphore Usage (Unix Masters)
With many process running on the master, it is important to review system semaphore usage and potentially raise the semaphore limits to ensure no semaphore constraints. The following are recommended values:
Recommended sem values:
300 307200 32 1024
More information on configuring these values can be found in the following article:
File Descriptors (Unix Masters)
If the operating system doesn't have enough file descriptors configured, performance will be negatively impacted. File descriptors should be set to a minimum of 8000 and monitored to ensure that value is sufficient on high performing systems. More information on this recommendation can be found here:
Disk performance will have a significant impact on the performance of the NetBackup master. It is required that the average service times for disks NetBackup resides remain lower than 20ms. The following article is a outlines the underlying disk options and requirements:
If disk performance is non-optimal it may be beneficial to move components of NetBackup to different disks to remove I/O contention. The following article outlines the disk layout options for improved performance:
Minor tuning to the threads currently allocated to Sybase can also have a large improvement on the system. To diagnose and adjust these issues, please see the following article: