It shouldn't be too hard to run a test with sequential reads that would max the controllers/bandwidth. 1.4 Gb/s is very low, we routinely hit 1gbs doing ndmp dumps from SATA disks on a 3020.

-----Original Message----- From: owner-toasters@mathworks.com [mailto:owner-toasters@mathworks.com] On Behalf Of Blake Golliher Sent: Wednesday, July 21, 2010 9:13 AM To: ifstat Cc: toasters@mathworks.com Subject: Re: 10gb Network Bandwidth concern

How busy are the CPUs and the disks? Is this more read or write intensive? There are a lot of factors that could be limiting you, to start with I'm pretty sure those controllers top out well before 20gbit is saturated.

Of course if you have a perfstat you can share that'd be an easy way to find out where you are stuck at.

-Blake

Typed with my thumbs!

On Jul 20, 2010, at 11:44 PM, ifstat hockey@spidernet.net wrote:

We have installed dual port 10Gbe (x1008a-r6) in our FAS3170A,

configured

them as a vif using lacp, with load balancing as mac.

What we are noticing is that we are only get 1.4Gbits bandwidth per

port.

Even though we should realy be achiving at least 20Gbits.

The filers connect to a 6509 CISCO switch, with each interface

connecting to

a different module for redundancy. The two interfaces from the filer

define

a port channel on the switch so that they reflect the vif on each

filer.

The tool we use to check on the throughput is called Solarwinds and

Orion.

This tool shows that: a. For netapp1, each interface is shown to work at 1.4101Gbits. The

vif is

shown to work at 2.8201Gbits. b. For netapp2, each interface is shown to work at 10Gbits. The vif is

shown

to work at 20Gbits, but we don't get more that 2.8201 Gbits

Is this normal or what are we misising? We have a case open (2001542643) is anyone for NetApp has access.

Thanks

View this message in context:

http://old.nabble.com/10gb-Network-Bandwidth-concern-tp29222397p29222397 .html

Sent from the Network Appliance - Toasters mailing list archive at

Nabble.com.

Please be advised that this email may contain confidential information. If you are not the intended recipient, please notify us by email by replying to the sender and delete this message. The sender disclaims that the content of this email constitutes an offer to enter into, or the acceptance of, any agreement; provided that the foregoing does not invalidate the binding effect of any digital or other electronic reproduction of a manual signature that is included in any attachment.

On that same note, it really takes a lot for the average host to push 1Gb/s, let alone 10Gb/s. Just remember, the data being pushed has to come from somewhere and it needs to go somewhere (depending on read or write). This also assumes that you have the appropriate amount of spindles in the storage aggregate in question to support this kind of I/O. I would say that before you start looking at the network, you may need to first qualify it as the bottleneck. It is more likely to be in other places. It is possible that when these are eliminated that the bottleneck could become the network, but you will have to build to that.

For an example of what it would take, I was able to get up to 10Gb/s with two FAS3170 controllers each with an aggregate of 44x 450GB disks in two RAID groups of 22 disks. The operation was a SnapMirror operation which eliminated host OS tweaks, etc.

Speaking of which, are the hosts optimized for this throughput? This typically entails tweaking their IP stacks and the respective protocol (ie. NFS tweaks for block sizes, etc).

Stetson Webster Professional Services Consultant Virtualization and Consolidation NCIE-SAN, NCIE-B&R, SCSN-E, VCP

NetApp 919.250.0052 Direct Phone stetson@netapp.com Learn more: http://www.imaginevirtuallyanything.com

-----Original Message----- From: Shane Garoutte [mailto:sgaroutte@gmail.com] Sent: Wednesday, July 21, 2010 11:12 AM To: ifstat Cc: toasters@mathworks.com Subject: Re: 10gb Network Bandwidth concern

Forgive for asking the obvious question, but the clients (servers) connected are using what speed of interface? Ie. Are you certain they can generate enough traffic to stress a 2x 10Ge VIF?

On Jul 20, 2010, at 11:44 PM, ifstat wrote:

We have installed dual port 10Gbe (x1008a-r6) in our FAS3170A,

configured

them as a vif using lacp, with load balancing as mac.

What we are noticing is that we are only get 1.4Gbits bandwidth per

port.

Even though we should realy be achiving at least 20Gbits.

The filers connect to a 6509 CISCO switch, with each interface

connecting to

a different module for redundancy. The two interfaces from the filer

define

a port channel on the switch so that they reflect the vif on each

filer.

The tool we use to check on the throughput is called Solarwinds and

Orion.

This tool shows that: a. For netapp1, each interface is shown to work at 1.4101Gbits. The

vif is

shown to work at 2.8201Gbits. b. For netapp2, each interface is shown to work at 10Gbits. The vif is

shown

to work at 20Gbits, but we don't get more that 2.8201 Gbits

Is this normal or what are we misising? We have a case open (2001542643) is anyone for NetApp has access.

Thanks

View this message in context:

http://old.nabble.com/10gb-Network-Bandwidth-concern-tp29222397p29222397 .html

Sent from the Network Appliance - Toasters mailing list archive at

Nabble.com.