Skip to main content
  • Bravura Security Fabric Documentation
  • Maintenance
  • Performance

Performance

After using Bravura Security Fabric for several years, the database can grow considerably and can affect performance and some important functions of the product.

Many performance improvement tasks require analysis of the data sets and stored procedures. It is highly recommended to purchase Bravura Security Professional Services.

Best practices

  • Ensure your server meets the minimum requirements:

  • Recommended database configuration:

    • Recovery mode - Simple

    • Snapshot Isolation - On

    • Create statistics - True

    • Update statistics - True

    • Update statistics asynch . - False

  • Store files:

    • On SSDs - only 50% of free space is required for performance execution without decreasing disk I/O.

    • On ext file systems - 40% free space is required. An ext3 file system is recommended over ext4 for not losing data during power failures; zfs is recommended for maximum security, at a minor speed reduction but increase in number of disks required.

    If files are stored on a Windows NTFS partition on a hard disk rather than an SSD, and if there is less than 2.5x (250%) the size of the files available in contiguous free space, that could lead to physical file fragmentation and increase disk I/O for the database.

  • Set up monitoring on all servers for available physical resources such as RAM, CPU cycles, disk space, and on very busy servers, sockets and handles.

    When any of these resources are close to depletion, especially at their peaks of load, provision more.

  • Do not schedule index rebuilding when large stored procedures like some of the psupdate ones are know to run, because it can lock/stall each other, leading to long-running stored procedures.

    Schedule index rebuilding after bulk processes like psupdate are completed.

  • For finite resources like sockets and handles, reduce the load on the server by adding more servers in replication and behind load balancers.

  • If the database server waits a long time for Network I/O and Buffer I/O, check with the owners of the hypervisor where the Bravura Security Fabric database node runs. Tweak those settings so that virtual machine gets more priority on those channels.

  • Monitor long-running stored procedures so you can detect ahead of time when some stored procedures are taking longer than they normally do; they tend to cause replication delays.

  • Navigate to Manage the system > Maintenance> Database replication on each node to monitor the queues.

  • To check what is using up resources in MSSQL, use the server's Activity Monitor in SSMS, or an alternative solution your DBA suggests to monitor the server's resources.

See Also

In this section: