techstyled
New Around Here
I'm trying to gauge the real world performance of our new NAS, a Scale Computing M4 cluster. I have a bunch of IOPS graphs and lots of other pretty stuff I don't understand (I'm the web developer) but I'm trying to get a more "real world" understanding of it's performance, especially in relation to our existing solutions (we have an EMC NS20 we are replacing with the Scale solution as well as some traditional Windows based file servers).
I'm most concerned about CIFS performance. We are using the native CIFS's of the Scale and we also have a virtual and a physical Windows box fronting CIFS's with an iSCSI connection to the Scale. Maybe a breakout of our solution options is warranted here:
I want to know which performs best over time (because the Scale has tier storage smarts i.e. caching).
I want to know this in terms I can understand. Perhaps Mbps or something? I don't "share" IOPS so I don't really know how to gauge results that show me in terms of that though I understand it can be helpful as a relative measurement when comparing solutions.
I realize for an exhaustive answer on the above you'd need to know a LOT more information about the type of information I'm sharing, how I'm accessing it and about 5,000 other pieces of information I probably don't know but I hope it will suffice to say that I'm mainly interested in using this as shared storage for a web farm so 4 or 5 other servers accessing the shared storage solution and I have no idea the type or manner of the access other than that it will DEFINITELY be CIFS.
I'm trying to find some easy to understand measurement process that's a notch above me copying a 4GB file around or such. My net admin has suggested IOZone but that seemed to be another IOPS type program. I don't want to become a storage expert, I just want to know that I'm using the best solution for shared storage for my web farm and I want to test it myself, not rely on my net admin (not that I don't trust him).
Anyone have any ideas?
I'm most concerned about CIFS performance. We are using the native CIFS's of the Scale and we also have a virtual and a physical Windows box fronting CIFS's with an iSCSI connection to the Scale. Maybe a breakout of our solution options is warranted here:
- Windows based file server w/ DAS
- EMC NS20 native CIFS
- Scale Computing M4 native CIFS
- Windows based file server using iSCSI connection to Scale
I want to know which performs best over time (because the Scale has tier storage smarts i.e. caching).
I want to know this in terms I can understand. Perhaps Mbps or something? I don't "share" IOPS so I don't really know how to gauge results that show me in terms of that though I understand it can be helpful as a relative measurement when comparing solutions.
I realize for an exhaustive answer on the above you'd need to know a LOT more information about the type of information I'm sharing, how I'm accessing it and about 5,000 other pieces of information I probably don't know but I hope it will suffice to say that I'm mainly interested in using this as shared storage for a web farm so 4 or 5 other servers accessing the shared storage solution and I have no idea the type or manner of the access other than that it will DEFINITELY be CIFS.
I'm trying to find some easy to understand measurement process that's a notch above me copying a 4GB file around or such. My net admin has suggested IOZone but that seemed to be another IOPS type program. I don't want to become a storage expert, I just want to know that I'm using the best solution for shared storage for my web farm and I want to test it myself, not rely on my net admin (not that I don't trust him).
Anyone have any ideas?
Last edited:
