Issues with 3-Node HA deployment
Posted: Thu Feb 13, 2014 3:52 pm
We installed Starwind Native SAN 3-Node (4TB) version 6.0.5713 back in December last year. Since then we have been facing very annoying issues to the point we are starting considering replacing this setup with a proper DELL Powervault SAN solution. We liked the idea of having redundancy with Starwind but the reality is that this is not working as we expected. Any how here's some background of our install:
- Starwind Native NAS running on 3 Dell Server R710 (Dual Xeon E5520, 72GB or RAM and Windows Server Standard 2012 R2)
- RAID 0 (4x 2TB NL-SAS HDDs) for performance (about 500 MB/s read/write performance)
- All 10Gb adapters and switches for Starwind (for both Synch channel and iSCSI traffic)
- 4TB HA device in Starwind with 1024MB of write-back caching (about 2.4 TB of actual data inside the HA device)
- 3-node cluster in Hyper V 2012 R2
Here are the issues:
1) The first issue we had was performance related. Even though when benchmarking the RAID 0 on each individual node we were getting 500 to 550 MB/s Read/Write speeds, when connecting to the Starwind Target we barely get around 200 MB/s.
2) Full sync speed is slow. The current configuration should allow for a full sync of 2.4TB happening in maximum 2.5 hours, however it takes between 8 and 14 hours.
3) Full Sync makes the system crawl! The VMs become extremely slow while a Full Sync is happening. If it's during a business day then all users complain of the slow performance.
4) The really worst issue we have is that any single time we reboot one of the 3 servers it requires a Full Sync! We thought if a server is shutdown properly then Fast Sync should happen. In the past month we had to reboot a server about twice a week, so 8 days in a month our users had to live with crawling speeds on the VMs.
We try contacting Starwind twice via email for support to no availability. Having paid over $5K I think we expected a more reliable product. We would really appreciate if we could get some help from the community on this regards.
Thanks!
- Starwind Native NAS running on 3 Dell Server R710 (Dual Xeon E5520, 72GB or RAM and Windows Server Standard 2012 R2)
- RAID 0 (4x 2TB NL-SAS HDDs) for performance (about 500 MB/s read/write performance)
- All 10Gb adapters and switches for Starwind (for both Synch channel and iSCSI traffic)
- 4TB HA device in Starwind with 1024MB of write-back caching (about 2.4 TB of actual data inside the HA device)
- 3-node cluster in Hyper V 2012 R2
Here are the issues:
1) The first issue we had was performance related. Even though when benchmarking the RAID 0 on each individual node we were getting 500 to 550 MB/s Read/Write speeds, when connecting to the Starwind Target we barely get around 200 MB/s.
2) Full sync speed is slow. The current configuration should allow for a full sync of 2.4TB happening in maximum 2.5 hours, however it takes between 8 and 14 hours.
3) Full Sync makes the system crawl! The VMs become extremely slow while a Full Sync is happening. If it's during a business day then all users complain of the slow performance.
4) The really worst issue we have is that any single time we reboot one of the 3 servers it requires a Full Sync! We thought if a server is shutdown properly then Fast Sync should happen. In the past month we had to reboot a server about twice a week, so 8 days in a month our users had to live with crawling speeds on the VMs.
We try contacting Starwind twice via email for support to no availability. Having paid over $5K I think we expected a more reliable product. We would really appreciate if we could get some help from the community on this regards.
Thanks!