Network performance monitoring

[from88]

There're some options like: perSONAR, OWAMP, Netnorad?. About the last one - it's a tool that was created by Facebook engineering to monitor their DC. The best thing about Netnorad is that the tool is able to randomize source port numbers, so the polarization problems wont occur. Dont know if perSONAR or OWAMP could to that. The downside of Netnorad is that it's just source code, or something like that - you need to to additional programing job to run it.

Maybe you guys have some recomendations of network performance monitoring ? Some othertools? Or some existing ones?
Thanks

[Reza Sharifi]

Try SolarWinds

https://www.solarwinds.com/network-performance-monitor?&CMP=KNC-TAD-GGL-SW_NA_X_PP_CPC_LD_EN_PBRDB_DWA-X-X_X_X_X-X-945268746_45682556783_g_c_+solarwinds%20+performance%20+monitoring-b~224542370339~&kwid=MS7wdzgm

HTH

[johnlloyd_13]

hi,

are you looking for free or paid/licensed monitoring tools?

aside from solarwinds npm, i also like PRTG and OpManager

https://www.paessler.com/prtg

https://www.manageengine.com/network-monitoring/

see useful link for testing opmanager:

http://pcnsewannabe.blogspot.com/2017/05/configuring-syslog-snmp-and-netflow-on.html

[Felipe A. Amaya]

You can also look at AKIPS.

[from88]

thank you,

i think i described not exactly enough.

Im looking specifically for Network Bandwidth monitoring system, which is able to do end-to-end probing. Usualy one node acts like traffic generator, and other - responder.

[Reza Sharifi]

For bandwidth and latency monitoring, you can use Nagios or Grafana.  They are both free but you need to do some development with the software to tune it for your environment.

HTH

[from88]

Hello, maybe i've described not correctly enough. What i need is not network monitoring system. I need Network performance tester for end-to-end network measurements. Something like perfSonar, or Netnorad. Have you any recommendations. The prerequisite is that the software would be able to randomize source ports to avoid loadbalancing-polarziation poblem where all traffic goes through one link.

[Mark Malone]

We use liveaction for that as we have IWAN in place , good tool dont think its cheap though , we use it track our pfr through IWAN all our qos traffic etc , right down to the packet and see where its going the flow , what it contains and you can push config make path adjustmemnts etc for pfr traffic over multiple links for load balancing , not sure if it can dynamically do it we had to set it all up but you can track the traffic from src - dst by port or dscp right through the global network the path its taking everything its doing

[from88]

thank you, maybe you know if it support something like source port number randomization - to avoid traffic measurement only on one link (polarization) ?

[Mark Malone]

it uses dscp values to avoid that and policy pusher on a csr router we use too

[Qunli Zhang]

For your requirement, iPerf could be the one of the tools you want, HTH.

Best regards,

PZ

[from88]

Thank you,

Can I ASK ?

How randomization DSCP values relates to load balancing ? Do DSCP values calculates into hash when load balancing process is active ?

1) Maybe Iperf can do port randomization, but we need some kind of system which do it always and gerenrates alerts, graphs and etc.

Thank you:)

[Mark Malone]

from my setup the DSCP values are set in the policy , so 2 routers on a remote site as an example 1 with internet vpn , 1 with mpls , if you had rsync engineering traffic we punt it accross the Internet circuit as its not critical and bulk data large amounts , if we have voice ef DSCP 46 we send it accross the MPLS , now this works with LA with a combination of each router having full Netflow and is graphed through the application so you can see each flow right down to the packet and utilisation etc , this last bit im not 100% sure but i can check it when im back in the office , the pfr/IWAN and LA know if one of the lines is under serious pressure and it will flip over to the other line regardles of DSCP when its in that state , there are probabaly multiple policys setup on the CSR checking against each other and LA , i need to check that i didnt install it i did the remote offices and border routers but not the CSR , i know from meetings this is how it works when it comes under serious pressure as it was designed that both circuits are active active at all times and being utilsed and load balancing but by default it will first follow the DSCP value and which path we have chosen for it but it does conatins some smarts behind it like that , and wont keep trying to bundle traffic down 1 circuit when its heavily utilised

From what i know so far of it and were not fully globally deployed yet were in the last phases i have never heard of any hash values being used , more policy based and the pfr and LA app working together to ensure full LB accross both circuits