The Latest Gartner® Magic Quadrant™Hyperconverged Infrastructure Software
The Latest Gartner® Magic Quadrant™Hyperconverged Infrastructure Software
StarWind iSCSI SAN Version 8.0 RC
Moderators: anton (staff), art (staff), Max (staff), Anatoly (staff)
-
Bohdan (staff)
- Staff
- Posts: 435
- Joined: Wed May 23, 2007 12:58 pm
Hello we are working on the issue in our testlab, I'll let you know about the results ASAP.
Hi, can anyone confirm if this degradation of performance (between local storage vs. storage presented via a starwind cluster) is to be expected in the current production version of starwind?
as I was getting 100k iops and 900MB/s off the local storage (5 drive SSD Array), I would have expected two servers presenting a disk subsystem with that i/o would in an ideal world, double our iops and MB/s (as we would be writing to the two servers in the cluster simultaneously (two 5 drive SSD arrays), and reading from them both.. etc).
as I was getting 100k iops and 900MB/s off the local storage (5 drive SSD Array), I would have expected two servers presenting a disk subsystem with that i/o would in an ideal world, double our iops and MB/s (as we would be writing to the two servers in the cluster simultaneously (two 5 drive SSD arrays), and reading from them both.. etc).
-
Bohdan (staff)
- Staff
- Posts: 435
- Joined: Wed May 23, 2007 12:58 pm
May I ask you about several quick tests?
1) On server1 create imagefile device with wb cache (imagefile size = wb cache size).
Connect it from server3 via 10gb paths, run the performance test 2+ times. Show us the performance values.
2) On server2 create imagefile device with wb cache (imagefile size = wb cache size).
Connect it from server3 via 10gb paths, run the performance test 2+ times. Show us the performance values.
3) Using server1 and server2, create HA device based on two imagefile devices with wb cache (imagefile size = wb cache size).
Connect it from server3 via 10gb paths, run the performance test 2+ times. Show us the performance values.
Thank you!
1) On server1 create imagefile device with wb cache (imagefile size = wb cache size).
Connect it from server3 via 10gb paths, run the performance test 2+ times. Show us the performance values.
2) On server2 create imagefile device with wb cache (imagefile size = wb cache size).
Connect it from server3 via 10gb paths, run the performance test 2+ times. Show us the performance values.
3) Using server1 and server2, create HA device based on two imagefile devices with wb cache (imagefile size = wb cache size).
Connect it from server3 via 10gb paths, run the performance test 2+ times. Show us the performance values.
Thank you!
-
Bohdan (staff)
- Staff
- Posts: 435
- Joined: Wed May 23, 2007 12:58 pm
Yes. The fix will be included in the next build. Thank you.jcrowe wrote:When removing a target that is still busy, the warning says "Target was not removed because busy.." This should probably be "Target was not removed because it is busy" instead.