Re: [c-nsp] 3750G vs. Nexus for a SAN
It is on the price list. $5300.. I have on in production and one on order for a customer.. Nice switch... Jim -Original Message- From: cisco-nsp-boun...@puck.nether.net [mailto:cisco-nsp-boun...@puck.nether.net] On Behalf Of Asbjorn Hojmark - Lists Sent: Monday, November 09, 2009 9:31 AM To: Brian Landers Cc: cisco-nsp@puck.nether.net Subject: Re: [c-nsp] 3750G vs. Nexus for a SAN On Mon, 9 Nov 2009 09:05:34 -0500, you wrote: [Cat 2350G] Doesn't appear to be in the pricing tool yet, though? Every order goes on NPH and needs to go through the BU for approval. Pricing is 'known, but not public'. -A ___ 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] 3750G vs. Nexus for a SAN
What version of IOS does it run ? Base version or lite version ? Wim Holemans Network Services University of Antwerp -Original Message- From: cisco-nsp-boun...@puck.nether.net [mailto:cisco-nsp-boun...@puck.nether.net] On Behalf Of Jim McBurnett Sent: vrijdag 13 november 2009 5:17 To: Asbjorn Hojmark - Lists; Brian Landers Cc: cisco-nsp@puck.nether.net Subject: Re: [c-nsp] 3750G vs. Nexus for a SAN It is on the price list. $5300.. I have on in production and one on order for a customer.. Nice switch... Jim -Original Message- From: cisco-nsp-boun...@puck.nether.net [mailto:cisco-nsp-boun...@puck.nether.net] On Behalf Of Asbjorn Hojmark - Lists Sent: Monday, November 09, 2009 9:31 AM To: Brian Landers Cc: cisco-nsp@puck.nether.net Subject: Re: [c-nsp] 3750G vs. Nexus for a SAN On Mon, 9 Nov 2009 09:05:34 -0500, you wrote: [Cat 2350G] Doesn't appear to be in the pricing tool yet, though? Every order goes on NPH and needs to go through the BU for approval. Pricing is 'known, but not public'. -A ___ 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/ ___ 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] 3750G vs. Nexus for a SAN
- Nick Hilliard n...@inex.ie wrote: Incidentally, if you're planning to use the N5K as a fancy 1G switch, note that the system will change the switching mode from cut-through to store-n-forward for GE ports; cut-through is only supported for 10G transceivers. This may matter for iSCSI. Looking at the specs an alternative would be a central 5010 and two 2148 FEX as top-of-rack 1G. Using up to 40 of the 1G ports and 4 x 10G for uplink to N5K would make it line-rate, right? Has anyone got any experiences with this setup? Arne ___ 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] 3750G vs. Nexus for a SAN
On Tuesday 10 November 2009 01:56:07 am Michael Balasko wrote: All that being said we bought the 5K's to do 10G distribution for our core so your mileage may vary depending on needs. To digress a little, we considered using the Nexus 5000 as a 10Gbps core aggregation switch, because it's way cheaper than the WS-X6704/8 line cards. But given that we'd be looking at adding more bandwidth in terms of N x 10Gbps, it made more sense to consider boxes that will scale to native 40Gbps and 100Gbps Ethernet interfaces. But, if a network is sure they'll never need anymore 10Gbps port density or large bandwidth to serve several downstream/upstream routers, the Nexus 5000 is definitely good value for 10Gbps Ethernet aggregation, current software issues aside, of course. Cheers, Mark. signature.asc Description: This is a digitally signed message part. ___ 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] 3750G vs. Nexus for a SAN
You can read about the architecture here: http://www.cisco.com/en/US/prod/collateral/switches/ps9441/ps9670/white_paper_c11-462176.html I'll give you my understanding of it -- I appreciate any corrections if I miss the mark on something. I don't know whether the packets are buffered on input or on output. Both. Each port has a set of 416 virtual output queues (on the 5020 -- don't know if this is true for the 5010 or if there are half as many). A VOQ is, essentially, a queue for each egress port. In other words, on ingress, a packet is put into one of 416 queues (52 egress ports * 8 queues -- one for each 802.1p CoS). Congestion on one egress port doesn't impact traffic destined for other ports. Internally, the packets are moved around at greater than 10Gb speed (+20%), so there is egress buffering as well. This allows multiple packets to be queued up and sent out at 10Gb rate without interruption and is also used for flow control buffering. per-port buffers...quite a bit smaller than on other products 480KB per port shared between per-CoS ingress and egress buffers. Most are assigned to ingress, but I don't know the ratio. There is also buffering on the fabric itself, though I'm not entirely sure what its impact is in this scenario (I think it's primarily just used as an optimization to increase throughput). James Nick Hilliard wrote: On 06/11/2009 21:35, Gert Doering wrote: Out of curiosity: how does it cut-through if it has to multiplex multiple ports, as in: packets coming in on port A and B and leaving on C? As soon as two packets overlap (time-wise) on A and B, you can't do cut-through... The switch has per-port buffers; from what i remember, quite a bit smaller than on other products, as the unit is cut-through. You also need these buffers when you're operating 1G ports in store-n-forward mode. I don't know whether the packets are buffered on input or on output. Nick ___ 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] 3750G vs. Nexus for a SAN
On Tue, Nov 10, 2009 at 4:02 AM, Jason Gurtz jasongu...@npumail.com wrote: Any reason why you wouldn't go for fcoe on nexus 5k? :) It does look like that is what the box is really for. To answer the question, it all depends on what SAN goes in. A lot of the newer stuff with better value is iSCSI only and eschews FC in any form. Well i'm not to sure on the better value point - I doubt it will be long before netapp and the likes pop a fcoe cna into their kit. Current prices of gen-2 cna's are not really any more expensive than a dual port 10ge card so why wouldn't you go fcoe? No ip, no tcp windows, no need to chew cpu on hosts no managing authentication Maybe I better question to ask is how does the nexus 5k fare against 49xx switch doing iSCSI? I don't think it would make much difference really, 5k will have less latency not that it really matters for iscsi :) ~JasonG ___ cisco-nsp mailing list cisco-...@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] 3750G vs. Nexus for a SAN
Hi, On Mon, Nov 09, 2009 at 07:55:43AM +0100, Gergely Antal wrote: Did you look at the c2350 also? http://www.cisco.com/en/US/products/ps10116/index.html The data sheet sounds very nice indeed. What I can't see from there is: - does it support flow-control? - how big and how flexible are its buffers? (as compared to 2950/2960/3750) - is there a redundant power suppy option? gert -- USENET is *not* the non-clickable part of WWW! //www.muc.de/~gert/ Gert Doering - Munich, Germany g...@greenie.muc.de fax: +49-89-35655025g...@net.informatik.tu-muenchen.de pgpkmc5UhHG2L.pgp Description: PGP signature ___ 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] 3750G vs. Nexus for a SAN
Gert Doering wrote: Hi, On Mon, Nov 09, 2009 at 07:55:43AM +0100, Gergely Antal wrote: Did you look at the c2350 also? http://www.cisco.com/en/US/products/ps10116/index.html The data sheet sounds very nice indeed. What I can't see from there is: - does it support flow-control? sh int t0/1 flowcontrol Port Send FlowControl Receive FlowControl RxPause TxPause adminoper adminoper - --- --- Te0/1 Unsupp. Unsupp. off off 0 0 - how big and how flexible are its buffers? how can i check this from cmd? (as compared to 2950/2960/3750) - is there a redundant power suppy option? it has redundant power supply's. signature.asc Description: OpenPGP digital signature ___ 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] 3750G vs. Nexus for a SAN
Hi, On Mon, Nov 09, 2009 at 10:30:07AM +0100, Gergely Antal wrote: On Mon, Nov 09, 2009 at 07:55:43AM +0100, Gergely Antal wrote: Did you look at the c2350 also? http://www.cisco.com/en/US/products/ps10116/index.html The data sheet sounds very nice indeed. What I can't see from there is: - does it support flow-control? sh int t0/1 flowcontrol Port Send FlowControl Receive FlowControl RxPause TxPause adminoper adminoper - --- --- Te0/1 Unsupp. Unsupp. off off 0 0 Hmmm. what about the Gig ports? - how big and how flexible are its buffers? how can i check this from cmd? I think you can't. At least on the other switches, I have not yet found a way to ask the device about its buffer details. (as compared to 2950/2960/3750) - is there a redundant power suppy option? it has redundant power supply's. It has? Cool. (That's not clearly visible from the data sheet). gert -- USENET is *not* the non-clickable part of WWW! //www.muc.de/~gert/ Gert Doering - Munich, Germany g...@greenie.muc.de fax: +49-89-35655025g...@net.informatik.tu-muenchen.de pgpFSYNzvpoGN.pgp Description: PGP signature ___ 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] 3750G vs. Nexus for a SAN
Gert Doering wrote: Hi, On Mon, Nov 09, 2009 at 10:30:07AM +0100, Gergely Antal wrote: On Mon, Nov 09, 2009 at 07:55:43AM +0100, Gergely Antal wrote: Did you look at the c2350 also? http://www.cisco.com/en/US/products/ps10116/index.html The data sheet sounds very nice indeed. What I can't see from there is: - does it support flow-control? sh int t0/1 flowcontrol Port Send FlowControl Receive FlowControl RxPause TxPause adminoper adminoper - --- --- Te0/1 Unsupp. Unsupp. off off 0 0 Hmmm. what about the Gig ports? the same - how big and how flexible are its buffers? how can i check this from cmd? I think you can't. At least on the other switches, I have not yet found a way to ask the device about its buffer details. (as compared to 2950/2960/3750) - is there a redundant power suppy option? it has redundant power supply's. It has? Cool. (That's not clearly visible from the data sheet). sorry i was misleading you.It has modular power and fan trays,but its not redundant. signature.asc Description: OpenPGP digital signature ___ 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] 3750G vs. Nexus for a SAN
On Mon, Nov 9, 2009 at 1:55 AM, Gergely Antal sk...@skoal.name wrote: Did you look at the c2350 also? http://www.cisco.com/en/US/products/ps10116/index.html Very interesting, indeed. Would be nice to see a POE version as well (to compete with the Force10 S50V), but as it seems to be positioned specifically as a data center switch, that doesn't seem likely. Doesn't appear to be in the pricing tool yet, though? -- Brian C Landers http://www.packetslave.com/ CCIE #23115 ___ 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] 3750G vs. Nexus for a SAN
On Mon, 9 Nov 2009 09:05:34 -0500, you wrote: [Cat 2350G] Doesn't appear to be in the pricing tool yet, though? Every order goes on NPH and needs to go through the BU for approval. Pricing is 'known, but not public'. -A ___ 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] 3750G vs. Nexus for a SAN
Any reason why you wouldn't go for fcoe on nexus 5k? :) It does look like that is what the box is really for. To answer the question, it all depends on what SAN goes in. A lot of the newer stuff with better value is iSCSI only and eschews FC in any form. Maybe I better question to ask is how does the nexus 5k fare against 49xx switch doing iSCSI? ~JasonG ___ 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] 3750G vs. Nexus for a SAN
I realize this is cisco-nsp, but does anyone have any opinions on the Force 10 S-series for top-of-rack? Especially for iSCSI SAN. I've long been frustrated with Cisco's lack of a cost-effective 48 ports of gigE with a 10ge uplink switch. I don't really *need* a $12,000 layer 3 switch (or two) at the top of every rack in my data center! Another thing we found when considering 1G w/ 10G uplinks and value is Fujitsu XG0448. ~JasonG ___ 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] 3750G vs. Nexus for a SAN
Brian Landers wrote: I realize this is cisco-nsp, but does anyone have any opinions on the Force 10 S-series for top-of-rack? Especially for iSCSI SAN. I've long been frustrated with Cisco's lack of a cost-effective 48 ports of gigE with a 10ge uplink switch. I don't really *need* a $12,000 layer 3 switch (or two) at the top of every rack in my data center! A HP ProCurve 6600-48G-4XG is a bit less and has 4x 10 gig and 48x 10/100/1000 ports. And they actually tell you the packet buffer size in their spec sheets. Never used this model personally though, but I have some other HP switches and I've been happy with them. The price difference and functionally equal (for my needs) that I'd seriously consider HP if they had complete IPv6 support. Cisco-nsp seems to be the mot active list of the *-nsp and having this list as a resource is valuable. ~Seth ___ 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] 3750G vs. Nexus for a SAN
I too can vouch for the 5K's not being ready for prime time. Here is a short list of the advanced features we are trying to use- -Disable the HTTP/HTTPS server onboard -NTP Authentication -ACL's for SNMP access -VTY ACL's -VTP passthrough - VTP packets WILL NOT pass through this switch. Please save the VTY argument is bad for someone else. As far as the Cisco litmus test of it forwards packets so it's working as designed it operates fine, but until the above mention issues are fixed, we can't in good conscience roll them into production to find the real bugs. We have piles of TAC cases open for this and we have screamed loud enough to be in direct contact with the 5K business unit product manager. The official answer is hurry up and wait. In order to fix these Cisco bugs we bought a pair of Brocade Turboiron 24's which are now our only non-Cisco piece of kit out of over 400 devices. All that being said we bought the 5K's to do 10G distribution for our core so your mileage may vary depending on needs. If it were done again right this second, I'd look at Arista Networks. We demo'd their gear way back and was impressed with the support folks and the willingness to respond to issues by cutting code instead of providing a workaround of none or don't use that feature. They couldn't do RPVST+ at the time and that's why we looked elsewhere. They say to do it today and based on some of the folks I know work there I'm inclined to believe them. Mike -Original Message- From: cisco-nsp-boun...@puck.nether.net [mailto:cisco-nsp-boun...@puck.nether.net] On Behalf Of Brian Landers Sent: Sunday, November 08, 2009 7:34 AM To: Andrew White Cc: cisco-nsp@puck.nether.net Subject: Re: [c-nsp] 3750G vs. Nexus for a SAN I realize this is cisco-nsp, but does anyone have any opinions on the Force 10 S-series for top-of-rack? Especially for iSCSI SAN. I've long been frustrated with Cisco's lack of a cost-effective 48 ports of gigE with a 10ge uplink switch. I don't really *need* a $12,000 layer 3 switch (or two) at the top of every rack in my data center! On Sun, Nov 8, 2009 at 6:49 AM, Andrew White adwh...@inchix.net wrote: Any reason why you wouldn't go for fcoe on nexus 5k? :) On Sat, Nov 7, 2009 at 6:26 AM, Jason Gurtz jasongu...@npumail.com wrote: Not sure that you want to go with Nexus at this point. Its got some really nice features, however we keep running into code bugs . Not just stuff that's obscure and shows up in certain situations but real show- stoppers like being unable to form port-channels with HP blade servers. ___ 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] 3750G vs. Nexus for a SAN
Any reason why you wouldn't go for fcoe on nexus 5k? :) On Sat, Nov 7, 2009 at 6:26 AM, Jason Gurtz jasongu...@npumail.com wrote: Not sure that you want to go with Nexus at this point. Its got some really nice features, however we keep running into code bugs . Not just stuff that's obscure and shows up in certain situations but real show- stoppers like being unable to form port-channels with HP blade servers. Interesting assessment and sorry to hear about the microsoftish experience. We're not intending to use blades (ESX Server 4 on a number of HP DL380G6 is likely) and would like to do cross-box etherchannels for redundancy. Jeff mentioned the 4948 of which the 10G version looks great since we're wanting to mirror the san off-site over fiber. There's still a chance that fiber channel will happen though it looks like that doesn't really make sense in this day and age. Here, vendors are pushing the MDS9124 box. Thanks for the responses so far. ~JasonG ___ cisco-nsp mailing list cisco-...@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] 3750G vs. Nexus for a SAN
I realize this is cisco-nsp, but does anyone have any opinions on the Force 10 S-series for top-of-rack? Especially for iSCSI SAN. I've long been frustrated with Cisco's lack of a cost-effective 48 ports of gigE with a 10ge uplink switch. I don't really *need* a $12,000 layer 3 switch (or two) at the top of every rack in my data center! On Sun, Nov 8, 2009 at 6:49 AM, Andrew White adwh...@inchix.net wrote: Any reason why you wouldn't go for fcoe on nexus 5k? :) On Sat, Nov 7, 2009 at 6:26 AM, Jason Gurtz jasongu...@npumail.com wrote: Not sure that you want to go with Nexus at this point. Its got some really nice features, however we keep running into code bugs . Not just stuff that's obscure and shows up in certain situations but real show- stoppers like being unable to form port-channels with HP blade servers. Interesting assessment and sorry to hear about the microsoftish experience. We're not intending to use blades (ESX Server 4 on a number of HP DL380G6 is likely) and would like to do cross-box etherchannels for redundancy. Jeff mentioned the 4948 of which the 10G version looks great since we're wanting to mirror the san off-site over fiber. There's still a chance that fiber channel will happen though it looks like that doesn't really make sense in this day and age. Here, vendors are pushing the MDS9124 box. Thanks for the responses so far. ~JasonG ___ 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/ -- Brian C Landers http://www.packetslave.com/ CCIE #23115 ___ 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] 3750G vs. Nexus for a SAN
Did you look at the c2350 also? http://www.cisco.com/en/US/products/ps10116/index.html Brian Landers wrote: I realize this is cisco-nsp, but does anyone have any opinions on the Force 10 S-series for top-of-rack? Especially for iSCSI SAN. I've long been frustrated with Cisco's lack of a cost-effective 48 ports of gigE with a 10ge uplink switch. I don't really *need* a $12,000 layer 3 switch (or two) at the top of every rack in my data center! On Sun, Nov 8, 2009 at 6:49 AM, Andrew White adwh...@inchix.net wrote: Any reason why you wouldn't go for fcoe on nexus 5k? :) On Sat, Nov 7, 2009 at 6:26 AM, Jason Gurtz jasongu...@npumail.com wrote: Not sure that you want to go with Nexus at this point. Its got some really nice features, however we keep running into code bugs . Not just stuff that's obscure and shows up in certain situations but real show- stoppers like being unable to form port-channels with HP blade servers. Interesting assessment and sorry to hear about the microsoftish experience. We're not intending to use blades (ESX Server 4 on a number of HP DL380G6 is likely) and would like to do cross-box etherchannels for redundancy. Jeff mentioned the 4948 of which the 10G version looks great since we're wanting to mirror the san off-site over fiber. There's still a chance that fiber channel will happen though it looks like that doesn't really make sense in this day and age. Here, vendors are pushing the MDS9124 box. Thanks for the responses so far. ~JasonG ___ 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/ signature.asc Description: OpenPGP digital signature ___ 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/