Juniper MX series RPM probes

[Mayur]

Hi,

How to get JItter value between two links using juniper RPM probes on JUniper MX series router.

which probe is use to get the Jitter vlaue with which destination port.

rg,

mayur

[ObaidDogar]

Hi,

I already asked this question & the reply i got was - Its not possible to get accurate jitter value using probe running on same router. You need external hardware/probe to measure jitter.

[Mayur]

Hi,

thanks for the reply.

Let me know which probe i have to use if i want to check jitter approx vlaues on juniper MX series router.

[crouchingbadger]

Hi Mayur,

Your results on the MX will be inconsistent and often delayed.

If you wish to experiment anyway then use the ICMP ping probe:

set services rpm probe cust test School1 probe-type icmp-ping
set services rpm probe cust test School1 target address 10.0.0.1
set services rpm probe cust test School1 probe-count 5
set services rpm probe cust test School1 probe-interval 5
set services rpm probe cust test School1 test-interval 10
set services rpm probe cust test School1 routing-instance school

It's a simple ICMP ping and will provide RTD and Jitter for each probe. You don't need to set a corresponding probe at the remote end. You can additionally set the DSCP bits, although with the response times you get there's little point.

Example response times:

• Using hardware timestamping on SRX650<- (via MX80) ->SRX650 the RTD is ~0.6ms
• Using software timestamping from the same MX80 -> SRX650 the RTD is ~4ms, going anywhere up to 60ms when the RE is too busy to respond or process the answer.

For accurate results, buy some tiny SRX branch routers (SRX210 works) or EX-switches and put the probes into hardware timestamp mode. You can then make them poll each other across your infrastructure. (You can put each probe in its own routing instance if you wish to direct traffic into a specific VRF or VLAN whilst only needing a default route in each end device.)

set services rpm probe cust test School1 probe-type icmp-ping
set services rpm probe cust test School1 target address 10.0.0.1
set services rpm probe cust test School1 probe-count 5
set services rpm probe cust test School1 probe-interval 5
set services rpm probe cust test School1 test-interval 10
set services rpm probe cust test School1 routing-instance school
set services rpm probe cust test School1 hardware-timestamp

Ben

[fan]

Hello,

I was trying to configure a probe that can give me some rtt data between 2 L3VPN's routing-instances.

Both RIs are configured correctly since I can do a PE ping between the two. However the probe doesn't work for the routing-instances:

lab@erda# show
probe probe1 {
    test test1 {
        probe-type icmp-ping;
        target address 192.168.239.58;
        probe-interval 1;
        test-interval 10;
        inactive: destination-port 65403;
        source-address 192.168.239.59;
        inactive: routing-instance rpm_fm;
        thresholds {
            rtt 10;
        }
        traps rtt-exceeded;
        inactive: destination-interface ge-0/3/0.0;
    }
}
probe-server {
    tcp {
        port 65400;
    }
    udp {
        port 65401;
    }
}

[edit services rpm]
lab@erda# run show services rpm probe-results
    Owner: probe1, Test: test1
    Target address: 192.168.239.58, Source address: 192.168.239.59, Probe type: icmp-ping, Test size: 1 probes
    Probe results:
      Response received, Tue Jul 17 15:41:13 2012, No hardware timestamps
      Rtt: 379 usec
    Results over current test:
      Probes sent: 1, Probes received: 1, Loss percentage: 0
      Measurement: Round trip time
        Samples: 1, Minimum: 379 usec, Maximum: 379 usec, Average: 379 usec, Peak to peak: 0 usec, Stddev: 0 usec, Sum: 379 usec
    Results over last test:
      Probes sent: 1, Probes received: 1, Loss percentage: 0
      Test completed on Tue Jul 17 15:41:13 2012
      Measurement: Round trip time
        Samples: 1, Minimum: 379 usec, Maximum: 379 usec, Average: 379 usec, Peak to peak: 0 usec, Stddev: 0 usec, Sum: 379 usec
    Results over all tests:
      Probes sent: 73, Probes received: 73, Loss percentage: 0
      Measurement: Round trip time
        Samples: 73, Minimum: 339 usec, Maximum: 425 usec, Average: 383 usec, Peak to peak: 86 usec, Stddev: 21 usec, Sum: 27970 usec

[edit services rpm]
lab@erda# activate probe probe1 test test1 routing-instance

[edit services rpm]
lab@erda# commit
commit complete

[edit services rpm]
lab@erda# run show services rpm probe-results                  
    Owner: probe1, Test: test1
    Target address: 192.168.239.58, Source address: 192.168.239.59, Probe type: icmp-ping
    Routing Instance Name: rpm_fm
    Test size: 1 probes
    Probe results:
      No route to target, Tue Jul 17 15:41:34 2012
    Results over current test:
      Probes sent: 1, Probes received: 0, Loss percentage: 100
    Results over last test:
      Probes sent: 1, Probes received: 0, Loss percentage: 100
    Results over all tests:
      Probes sent: 1, Probes received: 0, Loss percentage: 100

[edit services rpm]

[crouchingbadger]

Hi Fan,

I've had a quick look through your config. My suggestions:

• Unless you have a specific reason to then you should steer clear of probe-server configurations and just use the most basic methods.  
• Forget destination port because the RE should respond without any configuration on the other end (except RE filters).
• Forget destination interface and hardware-timestamping unless you're using a services PIC. 
• Your ping works without the routing-instance line. You then activate it and the ping doesn't work. You don't show the source IP addresses moving into the routing instance, so any replies will just get dropped.

Here's the minimum config, let me know how you get on:

set services rpm probe probe1 test test1 probe-type icmp-ping
set services rpm probe probe1 test test1 target address 192.168.239.58
set services rpm probe probe1 test test1 probe-interval 1
set services rpm probe probe1 test test1 test-interval 10
set services rpm probe probe1 test test1 routing-instance rpm_fm