high load in rb1200

Hi,

I’ve got a RB1200 router showing load from 60 to 100%. IT manages a fairly big amount of traffic (35 Mbps on average, 25kpps average, with peaks of 70 Mbps and 40kpps) but nothing over the specs.

There are a couple of singular things in this router:

  • it manages several VLANs in one interface
  • there is a VoIP system with avg 250 concurrent calls on average behind the router, connected to one VLAN (the load started after switching on this system)

FW version is 5.2 (I know I have to upgrade, but planning the downtime is not easy).

Any idea on the reason behind this high load? Suggestions?

What is using your CPU in /tool profile

Hi,

[admin@datacenter] > /tool profile duration=10 freeze-frame-interval=1 ;
NAME USAGE
ppp 0%
www 0.5%
ethernet 0.5%
console 0.5%
ssh 0.5%
firewall 9.5%
winbox 0%
management 1.5%
idle 12%
profiling 0.5%
queuing 2%
routing 0%
bridging 0%
unclassified 72.5%

:smiley:
Quite curious: “unclassified” :smiley:

unclassified means you have a really old RouterOS version, upgrade to v5.16
in old versions, some processes didn’t have names yet. they were all called “unclassified”

Hi,

I don’t know if it was a coincidence, the effect of a reboot or the firmware upgrade. I’ve upgraded to 5.16 and after the reboot the load is much lighter (20-30%, with peaks going up to 60%).

The output of /tool profile shows (right now is our valley hour, but load is significantly lower than yesterday at the same hour):

[admin@datacenter] > /tool profile
NAME CPU USAGE
www all 0%
console all 0.5%
firewall all 7.5%
management all 0%
idle all 83%
queuing all 9%

Thanks for the tip! It took 75 seconds to reboot and upgrade, but they were worth it! :smiley:
graph_cpu_daily.png

would you please put a screen shot of your tools/profile in your pick time,

i bought a 1200 today but i afraid to put it in action , my config is 30 Vlan in first 5 interfaces ,
Queueing - 1000 Full firewalling rules , NAT , PPTP , two IP sec tunnel , multihome BGP 3 neighbors and i think i have 2000 TPS ,

waiting for Cloud core 1036, any one has any idea or suggestion that this device can handle this amount of traffic ?

My config is much simpler, but was handling > 25k pps. I’ve switched to a RB1100 AHx2 and now the performance impact is much lower.

[admin@datacenter] > /tool profile
NAME CPU USAGE
snmp all 0%
ethernet all 4.2%
console all 0%
firewall all 14.7%
winbox all 0%
management all 0.5%
idle all 74.2%
profiling all 0%
queuing all 6.2%
unclassified all 0%

You can see the graphs below. In the monthly and yearly average it is easy to identify when the SIP server was started and when we changed the RB1200 for a RB1100AHx2 (a week ago).
cpu_load.png

Thankx Ediaz ,
i think i should test 1200 in action then changing to 1100AH2 if needed , Thank for your reply , it was very use full ,

Hi,

I’ve running my setup with an RB1100Ahx2 for some time now. Network performance has improved, but I think it still has place for improvement.

/tool profile shows

[admin@datacenter] > /tool profile
NAME CPU USAGE
snmp all 0%
ethernet all 3.7%
console all 0%
firewall all 10.7%
winbox all 0.2%
management all 0.2%
idle all 78%
profiling all 0%
queuing all 6.7%
unclassified all 0.2%

So I guess it should perform very well, as it is idle most of its time. However I suspect that when the number of packets increases (see graphs), performance suffers.

Most of the packets come from a sip server handling more than 500 concurrent calls.

Any idea on how to make the sip server not impact so severely the rest of the services?
graph_net_device_packet.png
graph_net_device.png
graph_cpu.png