One of the most essential tasks a SQL Server database administrator performs is taking database backups.  Having just a backup plan though is only part of the strategy, testing and validating the backup by restoring outside of a production environment and checking for corruption is equally important.

The frequency of backups is determined by how much data loss is acceptable.  How fast you can restore your backup will determine how much downtime you need before the production database is back online.  One of the most common excuses why database administrators do not test and validate their backups with restore, and check for corruption, is “I don’t have the time”.

The problem is that database backup, restore, and corruption testing can be tedious and put strain on infrastructure, leading to increased database maintenance completion times. However we believe having an incomplete backup strategy is a dangerous proposition.

FlashBlade Enables Holistic Backup Strategy

The Pure Storage FlashBlade delivers unprecedented performance in SQL Server environments to reduce backup and restore completion times. Its architecture efficiently utilizes flash for huge read and write bandwidth performance.  FlashBlade’s massively parallel, “tuned for everything” design delivers highest, consistent performance for advanced applications with NFS, S3/object, SMB, and HTTP protocols.

    

FlashBlade system with 15 blades, each blade designed with massively parallel architecture directly to the flash

SQL Server Database Backup to FlashBlade Target Results

The following backup test was conducted with only 7 blades installed in a single 4U chassis (MAX 15 blades).  More blades installed can deliver higher performance relative to the supporting environment, network, switch, etc.

• Optimized Test Configuration:
  • Seven Windows volumes (reader threads) for source SQL Server database files
  • Seven SQL Server backup files (writer threads)
  • Seven FlashBlade blades
  • Seven FlashBlade VIP (virtual IP address) mapped to Windows host volumes
• Database 3.2TB, data size 1.6TB (realistic data)
• Completion time 14 minutes 26 seconds
• Backup throughput 1.9GB/sec

SQL Server Database Restore from FlashBlade Source Results

The following restore test was conducted with only 7 blades installed in a single 4U chassis (MAX 15 blades).  More blades installed can deliver higher performance relative to the supporting environment, network, switch, etc.

• Optimized Test Configuration:
  • Seven Windows volumes (writer threads) for source SQL Server database files
  • Seven SQL Server backup files (reader threads)
  • Seven FlashBlade blades
  • Seven FlashBlade VIP (virtual IP address) mapped to Windows host volumes
• Database 3.2TB, data size 1.6TB (realistic data)
• Completion time 18 minutes 11 seconds
• Restore throughput 1.5GB/sec

Whether you are restoring your production SQL Server database, testing SQLServer database backups with restore, or refreshing a SQL Server based development environment the Pure Storage FlashBlade reduces the time required for these tedious tasks.