Have a customer using the p4300 SAN with amazingly poor performance. Have any thoughts?
Esxi 4.1u2 on 3 proliant dl580 g7’s with 8 nics:
2 for iScsi, 2 for vMotion, rest for vm and management
SAN is two sets of 4300 starter SANs which equates to 4 nodes, fully sas enterprise with 8 drives each, 450gb 6gb each in a network raid 10 configuration.
Storage is fully separate on it’s own iSCSI switch, all nodes and servers are physically on this HP ProCurve 2910al-48G switch for testing, eventually we will move it to multiple switches for redundancy but checking out performance is what the focus is on right now.
San nics are setup with ALB as per the recommended configuration and right now jumbo frames are off, however I have been getting mixed information on turning jumbo frames…thoughts?
Also all four nodes are setup in one cluster. Should that be modified to two clusters with two nodes each? What helps performance here?
There is a single fail over manager now, to help with quorum should we have any node failures…
Justin, the HP Lefthand page is blank…can you please check this out or send me the articles you published on this topic? Thank you very much!
Yeah, unfortunately i just haven’t gotten to populating all of the pages. Is there something that you are seeing that you need help with ?
Have a customer using the p4300 SAN with amazingly poor performance. Have any thoughts?
Esxi 4.1u2 on 3 proliant dl580 g7’s with 8 nics:
2 for iScsi, 2 for vMotion, rest for vm and management
SAN is two sets of 4300 starter SANs which equates to 4 nodes, fully sas enterprise with 8 drives each, 450gb 6gb each in a network raid 10 configuration.
Storage is fully separate on it’s own iSCSI switch, all nodes and servers are physically on this HP ProCurve 2910al-48G switch for testing, eventually we will move it to multiple switches for redundancy but checking out performance is what the focus is on right now.
San nics are setup with ALB as per the recommended configuration and right now jumbo frames are off, however I have been getting mixed information on turning jumbo frames…thoughts?
Also all four nodes are setup in one cluster. Should that be modified to two clusters with two nodes each? What helps performance here?
There is a single fail over manager now, to help with quorum should we have any node failures…
That’s about all I can think to mention…thanks!
what kind of IOps and Mbps are you seeing? I need to know what “poor performance” means to you.
The P4000 is still only capable of so many IOps.
Also it depends on the goal of the project as to whether it should be seperated into two groups of two or one big group of 4.