Re: [c-nsp] Egress QoS on FE links with less than 100Mbps speeds
Marian Ďurkovič wrote: Flowcontrol doesn't need to be QOS-aware in this scenario. On the switch side, it's enough if it supports plain RX flowcontrol i.e. flowcontrol receive [desired|on]. Then the wireless link can send pause frames to slow down the switch port automatically to the real bandwidth and output buffering / QOS configuration on the switch port is applied as expected. That is assuming that the switch honours QoS and continues to prioritise packets appropriately when it receives a PAUSE from the radio gear - more often than not this is not the case and all traffic in the egress buffers will be equally affected. I recall reading (maybe on this list) that the Nexium 5k or 7k supports QoS-aware flow control - but I wouldn't bet on it being included in low-end switches any time soon. (not that I'd complain if it was) Regards, Brad ___ 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] Egress QoS on FE links with less than 100Mbps speeds
On Fri, 18 Dec 2009 10:16:08 +1000, Brad Henshaw wrote Flow control comes with its own set of challenges however such as varied support across vendors and models and the fact that it's almost never QoS-aware in the kind of edge switches you're using. Flowcontrol doesn't need to be QOS-aware in this scenario. On the switch side, it's enough if it supports plain RX flowcontrol i.e. flowcontrol receive [desired|on]. Then the wireless link can send pause frames to slow down the switch port automatically to the real bandwidth and output buffering / QOS configuration on the switch port is applied as expected. With kind regards, M. ___ 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] Egress QoS on FE links with less than 100Mbps speeds
Wow thanks Daniel that did the trick on the 3750 platform! Here's a sample config in case anyone ever needs it: interface FastEthernet1/0/23 switchport trunk encapsulation dot1q switchport trunk allowed vlan 666-670 switchport mode trunk load-interval 30 srr-queue bandwidth share 1 25 35 40 srr-queue bandwidth shape 10 0 0 0 srr-queue bandwidth limit 80 priority-queue out mls qos trust dscp spanning-tree portfast Jose On 12/16/2009 10:04 AM, Bielawa, Daniel W. (NS) wrote: Hello, We had the same issue on couple of links. We solved it with the following command. The number on the end is a percentage of link speed in 1 percent increments. This was done on a 3750G running 12.2(44)SE6, this command might or might not work on other platforms. srr-queue bandwidth limit (10-90) Thank You Daniel Bielawa Network Engineer Liberty University Network Services Email: dwbiel...@liberty.edu Phone: 434-592-7987 -Original Message- From: cisco-nsp-boun...@puck.nether.net [mailto:cisco-nsp-boun...@puck.nether.net] On Behalf Of Lobo Sent: Wednesday, December 16, 2009 8:45 AM To: cisco-nsp@puck.nether.net Subject: [c-nsp] Egress QoS on FE links with less than 100Mbps speeds We're doing some Catalyst testing to roll out QoS on our Ethernet network and have come up against a hurdle. On most of our backbone links in a MAN, the actual bandwidth between one C/O to another C/O is not always 100Mbps. There are times when the link is only capable of hitting say 80Mbps (we're a wireless isp) or less. Since we have to use a FE port for this type of connection, do the switches believe that they have 100Mbps of bandwidth to play with when putting packets into the appropriate queues? I'm a bit confused as to how the switches work in this fashion. If I were using CAT5 cables or fiber this would be simple to understand as the bandwidth would be fixed. :) This is an example of a configuration on a 3550-24 that I'm using: interface FastEthernet0/x mls qos trust dscp wrr-queue bandwidth 40 35 25 1 wrr-queue cos-map 1 0 1 wrr-queue cos-map 2 2 wrr-queue cos-map 3 3 4 6 7 wrr-queue cos-map 4 5 priority-queue out ! The switches that we use are 2950, 3550, 3750 and 6524s. With MQC and layer 3 QoS, I would know how to fix this by simply using the bandwidth command on the physical interface and basing my output policy-map to use bandwidth percent for each class. Layer 2 QoS doesn't seem to work this way though. Any help would be appreciated. Thanks. Jose ___ 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] Egress QoS on FE links with less than 100Mbps speeds
Hi Peter. The reason why the radio only works at less than 100M is because that's all the bandwidth it has. This is licensed based wireless technology for point to point shots between buildings. Bandwidths can be anywhere from 18M to 400M depending on which frequency you use and radio brand. For that 400M radio we use Gig interfaces so we would need to use the bandwidth limit command to make sure that it only operates at 40% vs 100%. So for the 2950s and 3550s it looks like we may not have much wiggle room. My recommendation might be to upgrade those all to 3750s. :) Thanks. Jose On 12/16/2009 2:55 PM, Peter Rathlev wrote: On Wed, 2009-12-16 at 08:45 -0500, Lobo wrote: [...] There are times when the link is only capable of hitting say 80Mbps (we're a wireless isp) or less. Since we have to use a FE port for this type of connection, do the switches believe that they have 100Mbps of bandwidth to play with when putting packets into the appropriate queues? The interface will take packets from the output queue and send them as fast as it can, so as long as there are packets to be sent they will be sent at 100 mbps. I'm a bit confused as to how the switches work in this fashion. If I were using CAT5 cables or fiber this would be simple to understand as the bandwidth would be fixed. :) The interesting things happen in the box that converts from 100 mbps to something less, i.e. the wireless bridge. Why is it sometimes less than 100 mbps? Is it simple loss because of varying signal quality? Does the wireless bridge compensate for this loss by retransmitting at layer 1, meaning a little RTT variance and some lost bandwidth? Or does it just drop and let the overlying protocols handle this? (In short: how do you measure it? TCP throughput is not a reliable measurement.) About the switch: The WRR you configure (on a 3550) is Weighted Round Robin; it doesn't define anything relating to how much bandwidth there actually is, it just defines how many packets from each queue to serve to the interface tx ring in each turn. The important bit though is IMHO that you use the priority queueing. This means that queue 4 (CoS 5) will _always_ be sent first. This should minimise loss when traffic crosses the wireless bridge. The switches that we use are 2950, 3550, 3750 and 6524s. With MQC and layer 3 QoS, I would know how to fix this by simply using the bandwidth command on the physical interface and basing my output policy-map to use bandwidth percent for each class. Layer 2 QoS doesn't seem to work this way though. On the 3750 you can use what Daniel mentioned: srr-queue bandwidth limit. AFAIK this just uses a time divisioning on the interface and throws away unused timeslots. Bear in mind that if the wireless bridge has a very shallow queue this might not work very well. This command isn't available on the 2950 or 3550. And even though a few (10GE) ports one the 6500/7600 platform support SRR, you can't cap the interface as such like this. ___ 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] Egress QoS on FE links with less than 100Mbps speeds
On Thu, Dec 17, 2009 at 08:40:10AM -0500, Lobo wrote: Hi Peter. The reason why the radio only works at less than 100M is because that's all the bandwidth it has. This is licensed based wireless technology for point to point shots between buildings. Bandwidths can be anywhere from 18M to 400M depending on which frequency you use and radio brand. For that 400M radio we use Gig interfaces so we would need to use the bandwidth limit command to make sure that it only operates at 40% vs 100%. So for the 2950s and 3550s it looks like we may not have much wiggle room. My recommendation might be to upgrade those all to 3750s. :) In fact a properly implemented sub-rate service should use ethernet flowcontrol to signal real available bandwidth to the switch. With flowcontrol working, no such tweaks are necessary. With kind regards, M. ___ 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] Egress QoS on FE links with less than 100Mbps speeds
Lobo wrote: Wow thanks Daniel that did the trick on the 3750 platform! Here's a sample config in case anyone ever needs it: interface FastEthernet1/0/23 srr-queue bandwidth limit 80 Glad to hear that worked. Be aware that bandwidth limiting on these platforms is not an exact science - you'll always need to test to ensure you're getting what you expect - sometimes you may need to tweak the SRR-Queue buffer allocations and thresholds especially if traffic is bursty. The same goes for implementation of flow control. (with regard to the testing requirement, not the buffering) Flow control comes with its own set of challenges however such as varied support across vendors and models and the fact that it's almost never QoS-aware in the kind of edge switches you're using. Regards, Brad ___ 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] Egress QoS on FE links with less than 100Mbps speeds
Hello, We had the same issue on couple of links. We solved it with the following command. The number on the end is a percentage of link speed in 1 percent increments. This was done on a 3750G running 12.2(44)SE6, this command might or might not work on other platforms. srr-queue bandwidth limit (10-90) Thank You Daniel Bielawa Network Engineer Liberty University Network Services Email: dwbiel...@liberty.edu Phone: 434-592-7987 -Original Message- From: cisco-nsp-boun...@puck.nether.net [mailto:cisco-nsp-boun...@puck.nether.net] On Behalf Of Lobo Sent: Wednesday, December 16, 2009 8:45 AM To: cisco-nsp@puck.nether.net Subject: [c-nsp] Egress QoS on FE links with less than 100Mbps speeds We're doing some Catalyst testing to roll out QoS on our Ethernet network and have come up against a hurdle. On most of our backbone links in a MAN, the actual bandwidth between one C/O to another C/O is not always 100Mbps. There are times when the link is only capable of hitting say 80Mbps (we're a wireless isp) or less. Since we have to use a FE port for this type of connection, do the switches believe that they have 100Mbps of bandwidth to play with when putting packets into the appropriate queues? I'm a bit confused as to how the switches work in this fashion. If I were using CAT5 cables or fiber this would be simple to understand as the bandwidth would be fixed. :) This is an example of a configuration on a 3550-24 that I'm using: interface FastEthernet0/x mls qos trust dscp wrr-queue bandwidth 40 35 25 1 wrr-queue cos-map 1 0 1 wrr-queue cos-map 2 2 wrr-queue cos-map 3 3 4 6 7 wrr-queue cos-map 4 5 priority-queue out ! The switches that we use are 2950, 3550, 3750 and 6524s. With MQC and layer 3 QoS, I would know how to fix this by simply using the bandwidth command on the physical interface and basing my output policy-map to use bandwidth percent for each class. Layer 2 QoS doesn't seem to work this way though. Any help would be appreciated. Thanks. Jose ___ 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] Egress QoS on FE links with less than 100Mbps speeds
On Wed, 2009-12-16 at 08:45 -0500, Lobo wrote: [...] There are times when the link is only capable of hitting say 80Mbps (we're a wireless isp) or less. Since we have to use a FE port for this type of connection, do the switches believe that they have 100Mbps of bandwidth to play with when putting packets into the appropriate queues? The interface will take packets from the output queue and send them as fast as it can, so as long as there are packets to be sent they will be sent at 100 mbps. I'm a bit confused as to how the switches work in this fashion. If I were using CAT5 cables or fiber this would be simple to understand as the bandwidth would be fixed. :) The interesting things happen in the box that converts from 100 mbps to something less, i.e. the wireless bridge. Why is it sometimes less than 100 mbps? Is it simple loss because of varying signal quality? Does the wireless bridge compensate for this loss by retransmitting at layer 1, meaning a little RTT variance and some lost bandwidth? Or does it just drop and let the overlying protocols handle this? (In short: how do you measure it? TCP throughput is not a reliable measurement.) About the switch: The WRR you configure (on a 3550) is Weighted Round Robin; it doesn't define anything relating to how much bandwidth there actually is, it just defines how many packets from each queue to serve to the interface tx ring in each turn. The important bit though is IMHO that you use the priority queueing. This means that queue 4 (CoS 5) will _always_ be sent first. This should minimise loss when traffic crosses the wireless bridge. The switches that we use are 2950, 3550, 3750 and 6524s. With MQC and layer 3 QoS, I would know how to fix this by simply using the bandwidth command on the physical interface and basing my output policy-map to use bandwidth percent for each class. Layer 2 QoS doesn't seem to work this way though. On the 3750 you can use what Daniel mentioned: srr-queue bandwidth limit. AFAIK this just uses a time divisioning on the interface and throws away unused timeslots. Bear in mind that if the wireless bridge has a very shallow queue this might not work very well. This command isn't available on the 2950 or 3550. And even though a few (10GE) ports one the 6500/7600 platform support SRR, you can't cap the interface as such like this. -- Peter ___ 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/