[c-nsp] 3750E cluster replacement
We have a network based on a VSS with 20G channels to 3750E-24 clusters top-of-rack. We are seeing a lot of discards on the cluster which connects to our NetApp SANs. I suspect this is because of the small buffers in the 3750E switches and the growth of our traffic to the SAN, especially ISCI traffic. I'm considering replacing this cluster with something else, but I'm not sure what to put there. I read that 4900M have larger buffer and this would offer the needed mix of 1G en 10G ports but you can't cluster these switches and seen the importance of the connected devices, this is not really an option. Buffering on nexus 55xx seems also better and there you have the vPc possibility. Do you consider this the way to go or has anyone else a suggestion for a (clustered) device to replace this 3750E cluster ? Wim Holemans Netwerkdienst Universiteit Antwerpen Network Services University of Antwerp ___ cisco-nsp mailing list cisco-nsp@puck.nether.net https://puck.nether.net/mailman/listinfo/cisco-nsp archive at http://puck.nether.net/pipermail/cisco-nsp/
Re: [c-nsp] 3750E cluster replacement
Why's it important you maintain a cluster? You're absolutely correct, 3750's are weak ToR switches. I would go with the 5500 if you find yourself looking toward a wider nexus deployment in the next 18-36 months. On Tue, Jul 26, 2011 at 7:03 AM, Holemans Wim wim.holem...@ua.ac.be wrote: We have a network based on a VSS with 20G channels to 3750E-24 clusters top-of-rack. We are seeing a lot of discards on the cluster which connects to our NetApp SANs. I suspect this is because of the small buffers in the 3750E switches and the growth of our traffic to the SAN, especially ISCI traffic. I'm considering replacing this cluster with something else, but I'm not sure what to put there. I read that 4900M have larger buffer and this would offer the needed mix of 1G en 10G ports but you can't cluster these switches and seen the importance of the connected devices, this is not really an option. Buffering on nexus 55xx seems also better and there you have the vPc possibility. Do you consider this the way to go or has anyone else a suggestion for a (clustered) device to replace this 3750E cluster ? Wim Holemans Netwerkdienst Universiteit Antwerpen Network Services University of Antwerp ___ cisco-nsp mailing list cisco-nsp@puck.nether.net https://puck.nether.net/mailman/listinfo/cisco-nsp archive at http://puck.nether.net/pipermail/cisco-nsp/ ___ cisco-nsp mailing list cisco-nsp@puck.nether.net https://puck.nether.net/mailman/listinfo/cisco-nsp archive at http://puck.nether.net/pipermail/cisco-nsp/
Re: [c-nsp] 3750E cluster replacement
We use clusters to protect us from hardware failures ; all servers and SAN are dual connected to both switches. We have plans to install nexus in another server room, we could install 5500s in both and use them as interconnect (replacing the interconnects now made with 3750E). Wim Holemans Netwerkdienst Universiteit Antwerpen Network Services University of Antwerp From: chandler.bass...@gmail.com [mailto:chandler.bass...@gmail.com] On Behalf Of Chandler Bassett Sent: dinsdag 26 juli 2011 13:14 To: Holemans Wim Cc: cisco-nsp Subject: Re: [c-nsp] 3750E cluster replacement Why's it important you maintain a cluster? You're absolutely correct, 3750's are weak ToR switches. I would go with the 5500 if you find yourself looking toward a wider nexus deployment in the next 18-36 months. On Tue, Jul 26, 2011 at 7:03 AM, Holemans Wim wim.holem...@ua.ac.bemailto:wim.holem...@ua.ac.be wrote: We have a network based on a VSS with 20G channels to 3750E-24 clusters top-of-rack. We are seeing a lot of discards on the cluster which connects to our NetApp SANs. I suspect this is because of the small buffers in the 3750E switches and the growth of our traffic to the SAN, especially ISCI traffic. I'm considering replacing this cluster with something else, but I'm not sure what to put there. I read that 4900M have larger buffer and this would offer the needed mix of 1G en 10G ports but you can't cluster these switches and seen the importance of the connected devices, this is not really an option. Buffering on nexus 55xx seems also better and there you have the vPc possibility. Do you consider this the way to go or has anyone else a suggestion for a (clustered) device to replace this 3750E cluster ? Wim Holemans Netwerkdienst Universiteit Antwerpen Network Services University of Antwerp ___ cisco-nsp mailing list cisco-nsp@puck.nether.netmailto:cisco-nsp@puck.nether.net https://puck.nether.net/mailman/listinfo/cisco-nsp archive at http://puck.nether.net/pipermail/cisco-nsp/ ___ cisco-nsp mailing list cisco-nsp@puck.nether.net https://puck.nether.net/mailman/listinfo/cisco-nsp archive at http://puck.nether.net/pipermail/cisco-nsp/
Re: [c-nsp] 3750E cluster replacement
$quoted_author = Holemans Wim ; We use clusters to protect us from hardware failures ; all servers and SAN are dual connected to both switches. You don't need the clustering if you run active-backup. It's only LACP that requires a stack or virtual chassis. cheers Marty ___ cisco-nsp mailing list cisco-nsp@puck.nether.net https://puck.nether.net/mailman/listinfo/cisco-nsp archive at http://puck.nether.net/pipermail/cisco-nsp/
Re: [c-nsp] 3750E cluster replacement
We do run lacp on most of our server, nas, connections. We also need these 2G channels towards our SAN to accommodate for the accumulated ISCI traffic coming from different servers. 3750E also have only one power supply, so we cluster them and use port-channels to protect against hw/power failures. Even when replacing the 3750E with nexus 55xx (if needed combined with FEX) we intend to double them and have portchannels on both. Wim Holemans Netwerkdienst Universiteit Antwerpen Network Services University of Antwerp -Original Message- From: cisco-nsp-boun...@puck.nether.net [mailto:cisco-nsp-boun...@puck.nether.net] On Behalf Of Martin Barry Sent: dinsdag 26 juli 2011 14:12 To: cisco-nsp@puck.nether.net Subject: Re: [c-nsp] 3750E cluster replacement $quoted_author = Holemans Wim ; We use clusters to protect us from hardware failures ; all servers and SAN are dual connected to both switches. You don't need the clustering if you run active-backup. It's only LACP that requires a stack or virtual chassis. cheers Marty ___ cisco-nsp mailing list cisco-nsp@puck.nether.net https://puck.nether.net/mailman/listinfo/cisco-nsp archive at http://puck.nether.net/pipermail/cisco-nsp/ ___ cisco-nsp mailing list cisco-nsp@puck.nether.net https://puck.nether.net/mailman/listinfo/cisco-nsp archive at http://puck.nether.net/pipermail/cisco-nsp/