RARE user and assistance email list

[Tim Chown <>]

Hi Csaba,

> On 23 Aug 2022, at 12:30, mc36 <> wrote:
> 
> hi,
> 
> so it took me to figure out how to use the tool under linux and firefox but 
> finally i arrived to following url:
> https://timemap.geant.org/tm/gr/d/uuJsB3sGk/rpm-latency?orgId=1&var-src=lo0.mx1.ams.nl.geant.net&var-dst=lo0.mx1.ham.de.geant.net&viewPanel=2

I think those are Juniper RPM measurements rather than TWAMP, that’s another 
option on that platform to gather latency and from that deduce jitter.  It’s 
important for real-time applications like LoLa, where you have distributed 
orchestras/musicians trying to play in sync.

> well, i have good and bad news for you, let me start with the bad one:
> 
> so cisco calls this performance-measurements and they used in topologies 
> above 128,
> and this part is not yet there... but, for the proper bgp-ct operation i'll 
> need it,
> to finally handle the topic, and kaliraj is also pushing me to do so, 
> because what
> we can show off is somewhat a hack from the rare's point of view...
> long story short, i would ask you a bit of patience here,
> at least i have a plan on how to have it properly implemented...

The project has only run RWAMP on Juniper so far.  Experience with CXisco 
would be helpful.

Our perfSONAR toolkit supports TWAMP, as a client or server, so there’s that 
to test with or against too.
 
> regarding the twamp alone, the feature is there and it works against cisco 
> xr761,
> moreover, i also use it heavily to feed the regular topology database as a 
> metric/cost.

Aaah, I hadn’t thought of that.  So it’s running anyway… question then is 
feeding results into a tool like TimeMap.

> that is, i have some vms in the kifu cloud4edu openstack and the 
> freertr.org infra at
> arubacloud... as things change and they engineer the trafiic, i see the 
> change and i
> can easily select the current best from the below ones:
> aruba.cz - kifu
> aruba.de - kifu
> aruba.cz - aruba.de
> all the above multiplied if i go above ipv4 or ipv6
> 
> in the short run, the plan is to have a new knob to select the target 
> topology to feed...
> (but again, it's not yet there, i feed the topo#0, if configured to do 
> so...)
> 
> so basically this is what i have now:
> 
> mchome#show ipv4 lsrp 1 metric
> iface             router        name         peer           metric  gotmet  
> delay
> sdn2.157          10.10.10.199  player       10.1.1.249     9       9
> sdn2.170          10.10.10.20   nas          10.1.1.194     9       11
> sdn2.176          10.10.10.11   noti         10.1.1.173     9       9
> sdn2.182          10.10.10.199  player       10.1.1.125     9       9
> sdn2.186          10.10.10.20   nas          10.1.1.122     9       11
> sdn2.189          10.10.10.2    working      10.1.1.94      9       9
> sdn2.196          10.10.10.8    mediapc      10.1.1.74      9       9
> sdn2.199          10.10.10.5    safe         10.5.1.53      9       9
> sdn901            10.2.1.6      mchome-demo  10.2.1.6       9       10
> sdn905            10.1.11.12    wifi         10.1.1.50      9       10
> access39151817    10.10.10.4    parents      10.18.127.177  7       5       
> 7 7 7 8 8 7 6 6 6 7
> access2038349823  10.10.10.37   rare10Gcpe   10.18.127.167  3       10      
> 5 5 6 4 15 6 5 4 6 4
> access1228131172  10.10.10.26   p4deb        10.18.127.128  4       4       
> 5 4 5 5 5 5 4 6 5 4
> access952007535   10.10.10.4    parents      10.18.127.146  7       6       
> 7 8 9 6 17 9 8 7 9 6
> access1639998894  10.10.10.18   www          10.18.127.193  14      15      
> 15 15 15 15 15 15 16 16 14 15
> access949788031   10.10.10.180  vpn          10.18.127.213  34      34      
> 35 35 35 34 34 36 35 35 35 35
> access196268959   10.10.10.37   rare10Gcpe   10.18.127.204  4       10      
> 5 6 5 3 15 7 5 3 5 4
> access1870866491  10.10.10.26   p4deb        10.18.127.174  5       5       
> 19 5 16 5 4 14 4 5 5 5
> access1253156799  10.10.10.18   www          10.18.127.158  14      16      
> 15 15 16 15 29 16 15 17 15 14
> access1260980156  10.10.10.180  vpn          10.18.127.131  25      24      
> 24 24 27 25 24 24 23 24 23 24
> access212356525   10.10.10.38   meso         10.18.127.172  25      25      
> 25 26 34 25 33 25 28 28 26 35
> access430977934   10.10.10.38   meso         10.18.127.183  25      25      
> 26 26 26 27 27 27 27 38 25 27
> access353406112   10.10.10.28   rare100Glns  10.18.127.153  3       10      
> 4 4 6 4 4 5 4 5 5 5
> access1092126920  10.8.3.26     player       10.18.127.203  11111   33333   
> 71 31 34 30 47 30 30 33 34 30
> access1544029489  10.8.3.26     player       10.18.127.122  11111   33333   
> 25 29 26 31 21 41 35 33 33 36
> 
> mchome#
> 
> 
> here the metric column is what i'm currently advertising into the igp...
> the gotmet is what i see from the neighbor is advertising into the igp...
> (it's there to easily spot the misconfigurations and really not a cpu 
> consuming to have...:)
> and finally, the delay column, which i found useful and decided to use on 
> my core...
> 
> all the above works with the other link state igps also, not just the 
> freerouter
> homegrown one (lsrp.nop.hu) and you can select a lot more parameters if you 
> want one...
> that is, the rtt for example could come from other sources if you dont have 
> that much
> spare bandwidth, like me, over these regular internet links:
> 
> noti(cfg-if)#router isis4 1 dynamic-metric ?
>  disabled - forbid echo requests
>  icmpecho - icmp echo requests
>  twamp    - twamp echo requests
>  udpecho  - udp echo requests
> 
> noti(cfg-if)#router isis4 1 dynamic-algo ?
>  average - take average
>  dif-avg - take average of differences
>  dif-max - take highest of differences
>  dif-min - take lowest of differences
>  dif-sum - take summary of differences
>  maximum - take highest
>  minimum - take lowest
>  none    - nothing
>  summary - take summary
> 
> noti(cfg-if)#router isis4 1 dynamic-?
>  dynamic-algo            - calculation to do
>  dynamic-divisor         - divide result
>  dynamic-ignore          - ignore small differences
>  dynamic-maximum         - highest result
>  dynamic-metric          - dynamic peer metric
>  dynamic-minimum         - lowest result
>  dynamic-multiply        - multiply result
>  dynamic-size            - number of measurement
>  dynamic-skip-max        - discard big measures
>  dynamic-skip-min        - discard small measures
>  dynamic-time            - measurement interval
> 
> noti(cfg-if)#router isis4 1 dynamic-

The good news is better than the bad I think :)

This isn’t a priority, I just wondered how far off we’d be from something 
like TimeMap on the GEANT backbone being available for the GP4L network.

Tim

> 
> 
> br,
> cs
> 
> 
> 
> 
> 
> 
> On 8/23/22 12:51, Tim Chown wrote:
>> Hi,
>> We were talking this morning in the LoLa/TimeMap subtask about the work to 
>> date there being largely Juniper focused, as it   s something implemented 
>> on the GEANT Juniper core / backbone network.  You can see the latency and 
>> jitter measurements here with edutainment login - 
>> https://timemap.geant.org/tm/.  Measurements use TWAMP.
>> There was some talk of TWAMP on Cisco, but it made me wonder whether 
>> FreeRtr has TWAMP support, and if so how easy it would be to configure 
>> measurements on the GP4L backbone?  A very low priority question :)
>> Tim