Synchronization Process

[comdot]

Can someone explain to me how the full synchronization process works?

I am feeling like I'm not getting the performance I should be on full sync and I need to understand the process better to be able to confirm diagnosis or understand why its not performing as I would expect.

For example I'm doing a full sync as we speak and it will run at full speed 7-10gbps for ages, then it will slow down, sometimes down to around 2-500mbps for an hour, then it will speed back up again and so on..

I know that from a hardware point of view it's fine - as a test if I turn on smb on my sync channel (10GB Intel 540 x2) and copy an 800gb starwind img file over to my second node via smb it does it happily at 8-9Gbps constantly for the whole transfer and it takes about 20% of the time that it would take me to sync the same size file with Starwind. I've tried it with the sync set to top priority and with everything accessing via iSCSI turned off so it's not trying to service requests as well as sync yet it doesn't make any difference.

I can get full line speed with iperf, so NICs are setup ok, jumbo frames, offloads etc. Raid is quick enough to keep up as you can see from the SMB test so I'm struggling to understand why I'm not getting more speed out of starwind.

Any ideas??

[Anatoly (staff)]

We need to know the following information from you in order to provide a solution:
1. Operating system on servers participating in iSCSI SAN and client server OS
2. RAID array model, RAID level and stripe size used, caching mode of the array used to store HA images, Windows volume allocation unit size
3. NIC models, driver versions (driver manufacturer/release date) and NIC advanced settings (Jumbo Frames, iSCSI offload etc.)
4. Network scheme.
Also, there is a document for pre-production SAN benchmarking:
http://www.starwindsoftware.com/images/ ... _guide.pdf
And a list of advanced settings which should be implemented in order to gain higher performance in iSCSI environments:
http://www.starwindsoftware.com/forums/ ... t2293.html
http://www.starwindsoftware.com/forums/ ... t2296.html

Please provide me with the requested information and I will be able to further assist you in the troubleshooting

[comdot]

I'm happy to provide this information but please can you answer the question I actually asked.

Which was for a high level explanation of how the full synchronization process actually works.

----------------

Info as requested;

1. Server 2012 for 2 x iSCSI san host. Hosts are also hosting Hyper-V cluster. 256gb RAM in each and 2x Xeon E5-2670 8 cores.

2. RAID array is intel RMS25CB080 (equivalent of LSI MegaRAID 9271) with cachecade v2. RAID 10, stripe size 256k, write back caching with a supercap BBU. Has 1GB cache on card and also configured with 512gb of SSD cache on the front end that holds hot blocks, cache is the newer lsi type (cachecade v2) that is read and write. Array is made up of 14 x 1TB 10k drives. Windows alloc size is 4096.

3. Intel x540-T2 10GB adapter 1 port for sync on xover and 1 port for for iSCSI. 4 x Intel i350 1GBE teamed for guest access out. Jumbo frames is active and tested as are offloads. As I say I can max the adapters out with iperf. Drivers are intel ProSet 17.4.

4. as I say 1 x 10GB for sync 1 x 10gb for iSCSI traffic in each box then 4x 1GBE teamed to the outside world.

I did apply the recommended TCP-IP settings from your documents but unfortunately they actually made the performance worse in server 2012 so I reverted them back to the defaults. From what I've read I believe the server 2012 settings are much better out of the box for coping with high speed networking.

As I say I have removed the hardware as an issue as if I turn file and print on in the sync adapter and copy a large file via smb I can maintain line speed...

Thanks

[Anatoly (staff)]

May I ask you to send the latest log file to support@starwindsoftware.com and includ the reference to this topic in the Subject of email?

Thank you