Tek-Tips is the largest IT community on the Internet today!
Members share and learn making Tek-Tips Forums the best source of peer-reviewed technical information on the Internet!
-
Congratulations strongm on being selected by the Tek-Tips community for having the most helpful posts in the forums last week. Way to Go!
CISCO 2600 CHALLANGE GUYS!! 1
- Thread starter talkdev
- Start date
eurobadger
Technical User
You can set the bandwidth on the serial interface, using the bandwidth command:
bandwith <number-in-kbps>
Otherwise the default is T1.
bandwith <number-in-kbps>
Otherwise the default is T1.
-
1
- #3
Eurobadger is right in only one thing, that command will change your port to 128. It won't seggregate your users from whatever else you want. It takes the whole port down to 128 so the port is using less bandwidth all together. This isn't what you asked for, you asked if there was a way to segregate your users traffic from other traffic.
Or in other words you are asking about traffic management. My best suggestion is to use something other than FIFO (First-In-First-Out). If you want to get specific as to what protocol and upper layer port(snmp,smtp,icmp,etc.) you want to prioritize. You can do so by using Priority-lists lik PQ's (Priority Queuing) or CQ's(Customized Queuing). PQ's are lists that have four designated queues: High, Medium, Normal, and Low. Traffic can then be assigned to the queues based on protocol, port numbers, or other criteria. As you can guess packets in the high priority queue are transmitted first and so on. However, all traffic in the queues lower than High are not transmitted until all the High priority traffic is handled. Which can lead to queue starvation as the lower priority traffic waits to be serviced.
CQ is similar to PQ but uses a dfferent way of prioritizing. CQ can be set-up like PQ as far as what protocol, port, etc. However it has up to 16 different queues that can be used. Priority is established by defining how many bytes can be transmited from each queue in turn. Which is the equivalent of allocating bandwidth to a protocol. The transmission size of each queue is specified in bytes. Say for example, queue 1 is for ip it allows up to 1200 bytes to be queued then sent, then Queue 2 is set up for ICMP and allows for 700 bytes to be queued, and queue 3 is set up for smtp and it allows for 2000 bytes to be queued. CQ would send queue 1 then move to queue 2 then to queue 3 after which it round robbins back to queue 1. Which makes it a fairer queue process than PQ.
You then have WFQ (Weighted Fair-Queueing). Which places priority in Network Flows. WFQ classifies pacjets into conversations by flow; packets with the same source IP address, destination IP address, source TCP or UDP port, protocol, and type of service (ToS), will all belong in the same corresponding flow. Each flow will have its' own queue; when a packet is assigned a flow, it is placed into that corresponding queue. During periods of congestion, WFQ will allocate an equal share of the bandwidth to each active queue. Ensuring that packets with larger byte counts do not monopolize the bandwidth. This makes WFQ fairer than PQ or CQ, because it handles the problems inherent in queuing scemes that are first come first served based like the PQ and CQ are.
You can also use CBWFQ (Class-Based WFQ).
It allows packets to be assigned different queues that are based on input interfaces, IP addresses, or Protocols. You can configure a Class of Service for different traffic types. Up to 64 classes can be configured, providing different service policies for each flow.
You can get a better description of this from ciscos website.
Sorry for the length of this post.
Any questions feel free to contact me
tschouten@weltman.com
any flames use a post for it, I'm sure to see it.
Or in other words you are asking about traffic management. My best suggestion is to use something other than FIFO (First-In-First-Out). If you want to get specific as to what protocol and upper layer port(snmp,smtp,icmp,etc.) you want to prioritize. You can do so by using Priority-lists lik PQ's (Priority Queuing) or CQ's(Customized Queuing). PQ's are lists that have four designated queues: High, Medium, Normal, and Low. Traffic can then be assigned to the queues based on protocol, port numbers, or other criteria. As you can guess packets in the high priority queue are transmitted first and so on. However, all traffic in the queues lower than High are not transmitted until all the High priority traffic is handled. Which can lead to queue starvation as the lower priority traffic waits to be serviced.
CQ is similar to PQ but uses a dfferent way of prioritizing. CQ can be set-up like PQ as far as what protocol, port, etc. However it has up to 16 different queues that can be used. Priority is established by defining how many bytes can be transmited from each queue in turn. Which is the equivalent of allocating bandwidth to a protocol. The transmission size of each queue is specified in bytes. Say for example, queue 1 is for ip it allows up to 1200 bytes to be queued then sent, then Queue 2 is set up for ICMP and allows for 700 bytes to be queued, and queue 3 is set up for smtp and it allows for 2000 bytes to be queued. CQ would send queue 1 then move to queue 2 then to queue 3 after which it round robbins back to queue 1. Which makes it a fairer queue process than PQ.
You then have WFQ (Weighted Fair-Queueing). Which places priority in Network Flows. WFQ classifies pacjets into conversations by flow; packets with the same source IP address, destination IP address, source TCP or UDP port, protocol, and type of service (ToS), will all belong in the same corresponding flow. Each flow will have its' own queue; when a packet is assigned a flow, it is placed into that corresponding queue. During periods of congestion, WFQ will allocate an equal share of the bandwidth to each active queue. Ensuring that packets with larger byte counts do not monopolize the bandwidth. This makes WFQ fairer than PQ or CQ, because it handles the problems inherent in queuing scemes that are first come first served based like the PQ and CQ are.
You can also use CBWFQ (Class-Based WFQ).
It allows packets to be assigned different queues that are based on input interfaces, IP addresses, or Protocols. You can configure a Class of Service for different traffic types. Up to 64 classes can be configured, providing different service policies for each flow.
You can get a better description of this from ciscos website.
Sorry for the length of this post.
Any questions feel free to contact me
tschouten@weltman.com
any flames use a post for it, I'm sure to see it.
- Thread starter
- #5
fpippo is correct, the bandwidth statement has no affect on the speed of the port. It's simply used for calculating routing metrics.
To control bandwidth available to certain users or applications I would suggest the use of something like CheckPoints Floodgate package.
Chris.
************************
Chris Andrew, CCNA
chris@iproute.co.uk
************************
To control bandwidth available to certain users or applications I would suggest the use of something like CheckPoints Floodgate package.
Chris.
************************
Chris Andrew, CCNA
chris@iproute.co.uk
************************
The most up to date methods of traffic control use the Modular Quality of Service Command Line Interface commands (MQC) available from IOS Version 12.1(5) onwards. It really depends whether you want to "police" the traffic, i.e. this would involve dropping traffic in excess of the information rate (128kbps); or if you want to "shape" the traffic, which generally involves no data loss but introduces delays by buffering data. This is not a good idea for time-critical traffic.
These command structures have been created to replace Committed Access Rate (CAR) and Generic Traffic Shaping (GTS), which are still perfectly valid ways of controlling traffic.
However, there is no one single method of implementing a solution, owing to the complexity and variety of the command set and the variables involved. So .... a thorough read of the subject would help, or come on one of my QoS courses! Hope this helps,
Chas (CCSI)
These command structures have been created to replace Committed Access Rate (CAR) and Generic Traffic Shaping (GTS), which are still perfectly valid ways of controlling traffic.
However, there is no one single method of implementing a solution, owing to the complexity and variety of the command set and the variables involved. So .... a thorough read of the subject would help, or come on one of my QoS courses! Hope this helps,
Chas (CCSI)
About Custom queueing, ... When i specify configure a router to say, give 60 percent priority to say http and 20 percent to smtp etc.. is the traffic outgoing or incoming? This is important to me because i have a link to the internet that is 512 incoming and 256 outgoing.. i would like to know which traffic i'm dealing with..
NewbieGirl
Technical User
Back to eurobadger's posting. I read somewhere that "setting the bandwidth on an interface has nothing to do with the actual link speed. Rather, it provides the value that some routing protocols use in calculating routing metrics. The default bandwidth is 1.544mps, which is a T1 link". So what is the truth?
Thanks,
Danielle
Thanks,
Danielle
The truth is that the "bandwidth" statement does not affect the actual bandwidth of the link. As stated above, it's just for routing metrics.
Chris.
**********************
Chris Andrew, CCNA, CCSA
chris@iproute.co.uk
**********************
Chris.
**********************
Chris Andrew, CCNA, CCSA
chris@iproute.co.uk
**********************
- Status