Hello,
Filesystem backup performance is driving me crazy! Hope anyone can help? Your feedback is much appreciated!
Issue:
-----------------------------------------------
Backup of the servers mentioned below is taking ages. When I am lucky, this is all backed up within 48 hours. It's not possible to re-run a backup during weekdays. Storage performance is more than sufficient. We have not too many performance demanding applications running. To my surpise, a regula Windows file copy job is faster then a Data Protector driven backup. It's large file systems containing houndredhousends of files. This won't guarantee best performance, of course, but how can I ensure that backup is running as fast as possible? I am running ~4 disk agents per 2 TB LUN already and utilized TreeSizeFree to measure sizes in order to assign Disk Agents accordingly.
Below a detailed describtion of the environment:
Environment:
-----------------------------------------------
-- Clients
--- Fileserver 1 (3.5 TB) Many small files
--- Fileserver 2 (3.5 TB) Many small files
--- Fileserver 3 (4.5 TB) Many small files
--- Fileserver 4 (4.5 TB) Many small files
--- Veeam server (7.0 TB) Few large files
All virtual machines on different blade servers. VMs have 1 Gbit/s NICs. Each blade server has 2 Gbit/s uplink to 2 ethernet switches and each switch has 4 Gibt/s uplink to the core switches HP Data Protector is connected to.
Data resides on HP 3par / IBM XIV storage using 4 Gbit/s fibre channel connectivity per blade. Blade center has 2 or 3x8 Gbit/s fibre channel connectivity to SANs.
-- Backup system
--- Server 1 (Cell Server)
--- Server 2 (Media Agent)
---- MSL 4048 (2x 4 Gbit/s FC, 2x LTO5)
Both servers have 4x 1 Gbit/s running in eitherchannel, connected to core switches. Server running media agent has 2x 4 Gbit/s fibre Channel connectivity to fabric fibre channel switches (Server 1 <-> MSL 4048 using 2x LTO5).