Quality of Service (QoS) for multicast equipment in UDP

[cadumep12]

Hello experts, good morning!
I would like some assistance or ideas for a scenario I'm facing. I'm new to dealing with networks, especially CISCO. Here's the lab setup I have:
I have a Cisco switch (WS-C3560X-48) to be used with TV equipment, which will communicate via Multicast. A VLAN was configured in access mode on the ports and communication was validated, but with some freezing issues. To fix this, the idea of configuring QoS, IGMP-Carrier, and IGMP Snooping was suggested. Once I adjust the IP targeting via the modulator (connected equipment), it's advisable to link the SWITCH PORT, mapping the MODULATOR by Mbits and IPs, setting up filters per port on the SWITCH, for example: MODULATOR 1 - IPs XXXXXXXXXXXXXXX - 20 Mbits flow connected to port 1 of CISCO SWITCH, remembering the QoS application for UDP, which by default is in TCP, focusing on filtering by MODULATOR with attention to the IPs configured on them and also the transport MEGABITS that will be connected to the SWITCH ports to prevent the total flow from overwhelming all SWITCH ports.
Example: port 2 - 192.168.2.11 - Modulator 2 will communicate with port 5 - 192.168.2.12 - GATEWAY LAND 1.
Has anyone seen a similar scenario or have an idea of how to perform these configurations?

[cadumep12]

Continua dando erros no equipamento final, no modulador na porta 1

[cadumep12]

The strange thing is that the devices connected to the multicast interfaces on the switch are experiencing a lot of errors, even though the physical layer has been validated. Could this be related to the switch configuration? I heard that IGMP might help in these cases.

[cadumep12]

It keeps giving errors on the final equipment, on the modulator on port 1

[Joseph W. Doherty]

What errors are you (specifically) seeing?

From your prior posted stats for g0/11, compared to your most recently posted stats for g0/11, I see transmission counts have increased, but the drop count has not.  I also see the first queue's enqueued stats hasn't increase too.  So, if your video packets are being (correctly) directed to that queue, it appears they aren't being dropped or excessively delayed.

To further understand your video traffic, I would need to know entry port(s) and egress port(s), and it's correct marking; AF41, correct?  BTW, believed I noticed g0/11 didn't indicate seeing such a marking entering or exiting that port.(?)

[cadumep12]

On my switch (Sw), I have a TV modulator connected to port Gig0/1, configured with IP 192.168.2.10, which generates multicast group requests. On port Gig0/11, I have a channel server. The interfaces are currently configured as follows:

interface GigabitEthernet0/1
description 192.168.2.10 - Modulador 1
switchport access vlan 3070
switchport mode access
interface GigabitEthernet0/11
description 192.168.2.31 IRD CISCO D9800 1 
switchport access vlan 3070
switchport mode access
priority-queue out

The marking is in the policy:

policy-map VIDEO-POLICY
class VIDEO-CLASS
set dscp af41
class class-default
set dscp default
However, the marking has been removed from the interface for testing.

[Joseph W. Doherty]

To be fully clear, what port does enter on and what port does it exit on?

[cadumep12]

Claro, Gig 0/1 seria a entrada e Gig 0/11 seria a saída.

[Joseph W. Doherty]

---

@cadumep12 wrote:

Claro, Gig 0/1 seria a entrada e Gig 0/11 seria a saída.

---

"Of course, Gig 0/1 would be the input and Gig 0/11 would be the output."

Okay, in that case, again, g0/11 didn't show any additional interface drops.  But it does show packets being enqueued, so you do have transient congestion.  Live/real-time video doesn't like latency (or jitter).

Interestingly, G0/1 also shows packets have been enqueued, although no drops.

For your video, if it's performance is still suffering, direct you video packets to the first queue with PQ enabled.

If you still have a video performance problem, possibly it's due to other factors, from speed of multicast replication on the switch, to the multicast source device having generation issues.

[cadumep12]

"For your video, if it's performance is still suffering, direct you video packets to the first queue with PQ enabled."
How can I make this configuration?

[Joseph W. Doherty]

If default egress mapping:

Distribution1#show mls qos maps dscp-output-q
   Dscp-outputq-threshold map:
     d1 :d2  0    1    2    3    4    5    6    7    8    9
     ------------------------------------------------------------
     0 :  02-01 02-01 02-01 02-01 02-01 02-01 02-01 02-01 02-01 02-01
     1 :  02-01 02-01 02-01 02-01 02-01 02-01 03-01 03-01 03-01 03-01
     2 :  03-01 03-01 03-01 03-01 03-01 03-01 03-01 03-01 03-01 03-01
     3 :  03-01 03-01 04-01 04-01 04-01 04-01 04-01 04-01 04-01 04-01
     4 :  01-01 01-01 01-01 01-01 01-01 01-01 01-01 01-01 04-01 04-01
     5 :  04-01 04-01 04-01 04-01 04-01 04-01 04-01 04-01 04-01 04-01
     6 :  04-01 04-01 04-01 04-01

You want this traffic to be in Q1.  Above has default DSCP values to use Q1, values 40 through 47.

Assuming device doesn't tag the relevant traffic with necessary DSCP tag, on ingress port, and with ingress policy, assign DSCP and value between 40 and 47.

[cadumep12]

Sure, it seems that after adjusting the ingress interface and prioritizing with mls qos trust dscp, there was some improvement. It was suggested to update this switch, which is currently running version 12.2(55)SE3, to the recommended version 15. Do you know where I can find this firmware and patch? Model: WS-C3560X-48 SW Image: C3560E-UNIVERSALK9-M

[Joseph W. Doherty]

Newer IOS can be hard to find as I believe those switches are EoL and also believe you need a maintenance contract to download a newer IOS (if even available from Cisco).

The 12.2(55)SE# train, I believe, was a vary stable software train, especially  toward the end of the releases (any release 3 is often somewhat suspect for stability), also I recall the early 15.x versions weren't too good.

Our forum "expert" on those 3K software releases is @Leo Laohoo , who might also be able to suggest where to obtain a newer IOS.

[Leo Laohoo]

There are three universally-recognized "stable" IOS for the 3560E/X, 3750E/X and they are: 

• 12.2(55)SE train;
• 15.0(2)SE train; and
• 15.2(4)E train

How to legally download the files is the problem because Cisco has made it a point to inflict as much pain and punishment as possible by hiding ways to legally download the file(s) &/or removing the "Customers Without Service Contracts" section.  

Give this a try.  

1.  Find the complete filename of the file you want to download.
2.  Note the web page address
3.  Read "Cisco IOS and IOS XE Software Internet Key Exchange Version 1 Fragmentation Denial of Service Vulnerabilities" and scroll down to the Customers Without Service Contracts.  
4.  Contact Cisco TAC. 
5.  Provide the filename (Step 1), the web page address (Step 2) and the security vulnerability bulletin (Step 3).  

And good luck.