Veritas NetBackup™ Appliance Capacity Planning and Performance Tuning Guide
- About this Guide
- Section I. Capacity planning
- Analyzing your backup requirements
- Designing your backup system
- Section II. Best Practices
- Section III. Performance tuning
- Section IV. Quick reference to Capacity planning and Performance tuning
Setting NET_BUFFER_SZ to 0 on the NetBackup client for better WAN Optimization performance
NetBackup 7.x by default set TCP receive buffer size to 32032 bytes on client node for restore job. TCP receive buffer size determines the amount of inbound data buffered by TCP stack while either waiting to properly sequence missing frames or waiting for the receiving application to read the already sequenced data. The following example is from tar log on RHEL6 platform, it shows the network receive buffer size used by the client node when running restore:
05:05:50 (259.001) Setting network receive buffer size to 32032 bytes
NetBackup appliance has provided an integrated WAN optimization component which aims to improve the outbound network transfer rate over WAN, its performance is highly related to the TCP receive buffer size according to the formula of "TCP throughput = TCP receive window size / Round-trip delay time". The default behavior of NetBackup client has posed a limitation to WAN Optimization performance with fixed TCP window size.
To achieve better WAN Optimization performance for restore job, placing a zero(0) into the /usr/openv/netbackup/NET_BUFFER_SZ file on the client node, which disables the TCP receive buffer tuning made by NetBackup, and lets the Operating System choose the proper value for TCP receive buffer size.