I am experiencing a slower than anticipated initial sync or FULL sync times on my HA Targets.
The following are my hardware, OS and logical config.
Software Version: StarWind 5.5 Servers with RAID Controller: 3Ware 9690SA 8-Port SAS/SATA
Hard Drives: Seagate Barracuda ES.2 ST31000640SS 1TB 7200 RPM 16MB Cache
RAID-5: 5 x 1TB (above drives)
Physical LUN: 3.63 TB
HA Node OS: Windows 2k8 R2 (standard)
RAID-5 Partition: Basic Disk - GPT - NTFS - Primary
HA Target: 500GB - 2048 MB Write-Back Cache, 1000ms Expiry Period
Dedicated sync/mirror 1Gbps NIC: Full-Duplex, Jumbo Frames and Flow Control Enabled.
Dedicated switch with Jumbo Frames and Flow-Control Enabled for SYNC/Mirror
I set up a 500GB HA target but it looks like it is going to take somewhere between 36-40 hrs to complete the initial sync! I am assuming a FULL resync later on would take the same amount of time? I am able to get 35-40% NIC utilization when doing a FULL sync on a smaller HA target, so I know the NIC interface is performing at new full speed.
I know there is a lot of layer involved in designing and analyzing the overall I/O performance of a storage system.
My Questions are:
1) Does this seem normal for the hardware config I have?
2) Is there a design guide/Best-Practices guide that gives some guidelines?
24 hrs for a FULL sync on a 500GB target seems very long. There is not way I can survice a 24hrs of degraded performance if a FULL resync is needed.
When the HA Target is synced, Performance seems to be fast and stable.
Any help or insite would be great! Thanks
The Latest Gartner® Magic Quadrant™Hyperconverged Infrastructure Software