From: Pavel Bykov (slidersv@gmail.com)
Date: Wed Mar 25 2009 - 07:01:31 ART
The test you did is bad, and it does not tell you anything.
It is more complex to set up a good testing scenario.
To get good quantifiable results, get IXIA and produce large, continuous,
burstless UDP streams, just over the PIR rate.
Ping that requires reply is very wrong for this testing in many ways.
So in short: set up a new teting, where you have endpoints that send/receive
traffic that does not requires any replies/windowing/measuring to run.
Therefore, no TCP, no too large rates, etc.
On Wed, Mar 25, 2009 at 4:10 AM, Le Ba Duy Man
<manlbd.hcm@ct-in.com.vn>wrote:
> I changed the requirement a bit to *traffic that conform the guaranteed
> bandwidth 80kbps will be marked with MPLS EXP 3, traffic in excess the
> guaranteed rate but within the burst rate (800kps) should be marked with an
> MPLS EXP 1, traffic in excess of the burst rate should be drop*
> I labbed up, tested by ping 100 packet with size 2000 bytes with below
> result:
>
> *Option 1: CAR*
>
> Rack1R3#sho run int s1/2
>
> Building configuration...
>
>
>
> Current configuration : 304 bytes
>
> !
>
> interface Serial1/2
>
> ip address 10.1.13.3 255.255.255.0
>
> rate-limit input 80000 15000 15000 conform-action
> set-mpls-exp-imposition-transmit 3 exceed-action continue
>
> rate-limit input 800000 150000 150000 conform-action
> set-mpls-exp-imposition-transmit 1 exceed-action drop
>
> serial restart-delay 0
>
> Rack1R3#sho int s1/2 rate-limit
>
> Serial1/2
>
> Input
>
> matches: all traffic
>
> params: 80000 bps, 15000 limit, 15000 extended limit
>
> *conformed 145 packets*, 165800 bytes; action:
> set-mpls-exp-imposition-transmit 3
>
> exceeded 58 packets, 37252 bytes; action: continue
>
> last packet: 6464ms ago, current burst: 0 bytes
>
> last cleared 00:00:29 ago, conformed 44000 bps, exceeded 10000 bps
>
> matches: all traffic
>
> params: 800000 bps, 150000 limit, 150000 extended limit
>
> *conformed 58 packets*, 37252 bytes; action:
> set-mpls-exp-imposition-transmit 1
>
> exceeded 0 packets, 0 bytes; action: drop
>
> last packet: 10440ms ago, current burst: 0 bytes
>
> last cleared 00:00:29 ago, conformed 10000 bps, exceeded 0 bps
>
>
>
> *Option 2: Using MQC 1 rate 3 colors*
>
> Rack1R3#sho policy-map 2R3C
>
> Policy Map 1R3C
>
> Class class-default
>
> police cir 80000 bc 2500 be 25000
>
> conform-action set-mpls-exp-imposition-transmit 3
>
> exceed-action set-mpls-exp-imposition-transmit 1
>
> violate-action drop
>
> Rack1R3#sho policy-map int s1/2
>
> Serial1/2
>
>
>
> Service-policy input: 1R3C
>
>
>
> Class-map: class-default (match-any)
>
> 203 packets, 203052 bytes
>
> 5 minute offered rate 9000 bps, drop rate 1000 bps
>
> Match: any
>
> police:
>
> cir 80000 bps, bc 2500 bytes, be 25000 bytes
>
> *conformed 119 packets*, 151196 bytes; actions:
>
> set-mpls-exp-imposition-transmit 3
>
> *exceeded 82 packets*, 49828 bytes; actions:
>
> set-mpls-exp-imposition-transmit 1
>
> *violated 2 packets*, 2028 bytes; actions:
>
> drop
>
> conformed 8000 bps, exceed 6000 bps, violate 1000 bps
>
>
>
> *Option 3: Using MQC 2 rates 3 colors*
>
> Rack1R3#sho policy-map 2R3C
>
> Policy Map 2R3C
>
> Class class-default
>
> police cir 80000 bc 2500 pir 800000 be 25000
>
> conform-action set-mpls-exp-imposition-transmit 3
>
> exceed-action set-mpls-exp-imposition-transmit 1
>
> violate-action drop
>
> Rack1R3#sho policy-map int s1/2
>
> Serial1/2
>
>
>
> Service-policy input: 2R3C
>
>
>
> Class-map: class-default (match-any)
>
> 202 packets, 202968 bytes
>
> 5 minute offered rate 7000 bps, drop rate 0 bps
>
> Match: any
>
> police:
>
> cir 80000 bps, bc 2500 bytes
>
> pir 800000 bps, be 25000 bytes
>
> *conformed 115 packets*, 148560 bytes; actions:
>
> set-mpls-exp-imposition-transmit 3
>
> *exceeded 87 packets*, 54408 bytes; actions:
>
> set-mpls-exp-imposition-transmit 1
>
> violated 0 packets, 0 bytes; actions:
>
> drop
>
> conformed 5000 bps, exceed 2000 bps, violate 0 bps
>
>
>
> So Option 2 & 3 seems producing the same result whereas Option 1 does
> differently. Can anyone share thoughts about this?
>
> BTW, I have two question regarding Bc, Be value:
>
> - With CAR: Bc (bits) = CIRx1.5 , Be=2xBc. Is it correct?
>
> - How can we calculate the Bc, Be value for 1 rate, 3 color policing?
>
> Thanks
> ------------------------------
>
> *From:* Pavel Bykov [mailto:slidersv@gmail.com]
> *Sent:* Wednesday, March 25, 2009 8:11 AM
> *To:* Le Ba Duy Man
> *Cc:* ccielab@groupstudy.com; rintrum@gmail.com
> *Subject:* Re: QOS: Policing
>
>
>
> Think about the requirements.
> What is being asked? The thing that should pop into your mind immediately
> is two continuous rates.
> Therefore, only 2R3C (two rate three color policer) will do the job.
>
> Option 2 is therefore out of the question.
>
> Option 1 and 3 are similar. 1 is configured using legacy CAR and 3 is
> configured using MQC and CIR/PIR combination with proper actions.
>
> But option 1 is tiny bit less precise then option 3. So Option 3 is he
> best.
> Also, in practice, 3 would be preferred, because we don't want use legacy
> unless there is a good reason, and I don't see this reason from here.
>
>
> On Sun, Mar 22, 2009 at 9:36 AM, Le Ba Duy Man <manlbd.hcm@ct-in.com.vn>
> wrote:
>
> Hi group,
>
>
>
> I have the following scenario: Configure the router that traffic coming in
> that conforms to the guaranteed banwidth (2.5Mbps) is marked with an MPSL
> EXP 1. Traffic in excess the guaranteed rate but within the burst rate
> (7Mbps) should be marked with an MPLS EXP 0. Traffic in excess of the burst
> rate should be drop.
>
> I have three possile solution for this
>
> 1. Using CAR
>
> interface FastEthernet0/0
>
> rate-limit input 2496000 468750 468750 conform-action
> set-mpls-exp-transmit
> 1 exceed-action continue
>
> rate-limit input 7000000 1312500 1312500 conform-action
> set-mpls-exp-transmit 0 exceed-action drop
>
>
>
> 2. Using MQC one rate, three color policing
>
> policy-map 1RATE_3COLOR
>
> class class-default
>
> police cir 2500000 bc 39062 be 109375
>
> conform-action set-mpls-exp-transmit 1
>
> exceed-action set-mpls-exp-transmit 0
>
> violate-action drop
>
>
>
> 3. Using MQC two rate, three color policing
>
> policy-map 2RATE_3COLOR
>
> class class-default
>
> police cir 2500000 pir 7000000
>
> conform-action set-mpls-exp-transmit 1
>
> exceed-action set-mpls-exp-transmit 0
>
> violate-action drop
>
>
>
> Which one is the most correct and why?
>
> Thanks,
>
> Rin
>
>
> Blogs and organic groups at http://www.ccie.net
>
> _______________________________________________________________________
> Subscription information may be found at:
> http://www.groupstudy.com/list/CCIELab.html
>
>
>
>
>
>
>
>
>
> --
> Pavel Bykov
> ----------------
> Don't forget to help stopping the braindumps, use of which reduces value of
> your certifications. Sign the petition at http://www.stopbraindumps.com/
>
-- Pavel Bykov ---------------- Don't forget to help stopping the braindumps, use of which reduces value of your certifications. Sign the petition at http://www.stopbraindumps.com/Blogs and organic groups at http://www.ccie.net
This archive was generated by hypermail 2.1.4 : Mon Apr 06 2009 - 06:44:07 ART