Veritas NetBackup™ Appliance Capacity Planning and Performance Tuning Guide

Last Published:
Product(s): Appliances (3.0)
Platform: NetBackup Appliance OS
  1. About this Guide
    1.  
      About this guide
    2.  
      About the intended audience
    3.  
      About the NetBackup appliance documentation
  2. Section I. Capacity planning
    1. Analyzing your backup requirements
      1. Analyzing your backup requirements
        1.  
          What do you want to backup?
        2.  
          How much data do you want to back up?
        3.  
          When should the backup take place?
        4.  
          What is the retention period?
        5. Record your backup requirements
          1.  
            Template to record computer system information
          2.  
            Template to record database information
          3.  
            Template to record application server information
    2. Designing your backup system
      1.  
        Addressing use cases of backup systems for enterprises
      2.  
        Addressing use cases of backup systems for remote or branch offices
      3. About NetBackup appliances
        1.  
          About the Master Server role
        2.  
          About the Media Server role
      4. Selecting new appliances
        1.  
          Selecting a media server
        2.  
          Selecting a master server
  3. Section II. Best Practices
    1. About the best practices section
      1.  
        About best practices
      2.  
        References to maintenance sites
    2. About implementing deduplication solutions
      1.  
        About implementing deduplication solutions
      2.  
        General recommendations
      3.  
        Oracle
      4.  
        Microsoft SQL
      5.  
        DB2
      6.  
        Sybase
      7.  
        Lotus Notes
    3. Network consideration
      1.  
        About network considerations
      2.  
        About Fibre Channel connectivity
      3. About SAN zone configurations
        1.  
          About zoning the SAN for NetBackup appliances
      4.  
        Validating network bandwidth
    4. Storage configuration
      1.  
        About storage configuration
      2. About configuring a shared storage pool
        1.  
          Calculating the basic stream count for backups
      3. About moving a storage partition for better performance
        1. Moving a partition
          1.  
            Move <partition> dialog
        2.  
          Moving a partition using the NetBackup Appliance Shell Menu
        3.  
          Moving the MSDP partition from a base disk to an expansion disk for optimum performance
    5. Generic best practices
      1.  
        Generic best practices
      2.  
        About Notification settings
      3.  
        About IPMI configuration
      4.  
        Disaster recovery best practices
      5.  
        Job performance
      6.  
        Architecture
      7.  
        NetBackup Catalog Backup
      8.  
        Patching with the SLP (storage lifecycle policies)
      9.  
        VMware backups using appliances
      10.  
        Improving NetBackup Appliance restore performance
  4. Section III. Performance tuning
    1. Role-based Performance Monitoring
      1.  
        Role-based configuration as a factor affecting performance
      2.  
        What affects the performance of a Master Server?
      3.  
        What affects the performance of a media server (MSDP)?
    2. Optimize network settings and improve performance
      1.  
        Optimize network settings and improve performance
      2.  
        SAN Fiber Channel setup
      3.  
        Network bonding
      4.  
        VMware VADP
      5.  
        Jumbo frame implementation for increased MTU
    3. Storage configurations
      1.  
        Storage configurations
      2.  
        Deduplication disk I/O and RAID level settings
      3. RAID Controller operations
        1.  
          RAID Controller commands
      4.  
        Deduplication load balancing
      5.  
        Storage Lifecycle Policies
      6.  
        Auto Image Replication (AIR)
      7.  
        AdvancedDisk settings
      8.  
        Tape Out operations
    4. NetBackup Appliance tuning procedures and performance monitoring
      1. About diagnosing performance issues
        1.  
          About CPU monitoring and tuning
        2.  
          About memory monitoring and tuning
        3.  
          About network monitoring and tuning
        4.  
          About I/O monitoring and tuning
        5.  
          General guidelines to spot a resource bottleneck
      2. About performance tuning practices
        1.  
          About I/O performance tuning
        2.  
          About Oracle backup and restore performance tuning
        3.  
          Setting NET_BUFFER_SZ to 0 on the NetBackup client for better WAN Optimization performance
      3. About tuning procedures and performance monitoring
        1. NetBackup Client performance
          1.  
            Using nbperfchk to validate disk performance on NetBackup Appliances
        2.  
          Other Performance Monitoring commands
  5. Section IV. Quick reference to Capacity planning and Performance tuning
    1. Capacity Planning checklist
      1.  
        Checklist for Capacity planning
      2.  
        Capacity measuring worksheet
    2. Best practices checklist
      1.  
        Best practices checklist
    3. How to monitor the performance of your appliance
      1.  
        About monitoring the performance of your appliance
      2.  
        Performance Monitoring matrix

General guidelines to spot a resource bottleneck

Identifying performance problem is not trivial, however, the following guideline can help quickly spot a potential problem:

  • CPU may be a bottleneck if us + sy from vmstat output is consistently over 85% and/or the r column is more than four times the number of CPU cores on the system

  • Memory may be a bottleneck if the value of "si" and "so" columns is consistently high in vmstat output (for example over 1000).

  • I/O subsystem may be a bottleneck if the await column of VxVM device show greater than 25(ms). Another way to detect I/O bottleneck is continue increasing the load with an I/O intensive workload. When the throughput stop increases but the await time increases substantially, then it is very likely that the I/O subsystem is saturated.

  • Network may be a bottleneck, if the sum of "in" + "out" is close to 1.2 GB/sec from ifstat output on a 10 Gbps or 120 MB/s on a 1 Gbps NIC.

For the first three items, consider lowering the batch size of concurrent jobs. Overall throughput may be better than overstressing the system because excessive jobs will most likely be on an internal queue somewhere waiting for resources to free up. For network bottlenecks, changing the bonding mode or adding more NIC ports should alleviate the problem.