hello
I have an interconnection through a vpn. When I activate the links in nv2 the vpn latency is increased while times wan network remains the same. if i activate nstreme in this links the latency of vpn and wan is exactly the same. nv2 and nstreme is the diference for increase latency.
have nv2 a problem for l2tp sessions ?
No. Nv2 just have higher ping and speed.
We’ve similar problem without VPN.
When there are no bandwidth in use the latency has spike and jitter is unstable.
I attach some example on a CPE
Ping WITHOUT traffic in use
See the spikes and the unstability
ping XX.XX.21.129
HOST SIZE TTL TIME STATUS
XX.XX.21.129 56 64 2ms
XX.XX.21.129 56 64 34ms
XX.XX.21.129 56 64 7ms
XX.XX.21.129 56 64 12ms
XX.XX.21.129 56 64 12ms
XX.XX.21.129 56 64 51ms
XX.XX.21.129 56 64 21ms
XX.XX.21.129 56 64 13ms
XX.XX.21.129 56 64 19ms
XX.XX.21.129 56 64 5ms
XX.XX.21.129 56 64 36ms
XX.XX.21.129 56 64 8ms
XX.XX.21.129 56 64 40ms
XX.XX.21.129 56 64 13ms
XX.XX.21.129 56 64 18ms
XX.XX.21.129 56 64 2ms
XX.XX.21.129 56 64 5ms
XX.XX.21.129 56 64 18ms
XX.XX.21.129 56 64 18ms
XX.XX.21.129 56 64 11ms
sent=20 received=20 packet-loss=0% min-rtt=2ms avg-rtt=17ms max-rtt=51ms
Same CPE: ping with a 200Kbps busy
(See the stability)
ping XX.XX.21.129
HOST SIZE TTL TIME STATUS
XX.XX.21.129 56 64 4ms
XX.XX.21.129 56 64 3ms
XX.XX.21.129 56 64 4ms
XX.XX.21.129 56 64 12ms
XX.XX.21.129 56 64 2ms
XX.XX.21.129 56 64 3ms
XX.XX.21.129 56 64 4ms
XX.XX.21.129 56 64 17ms
XX.XX.21.129 56 64 3ms
XX.XX.21.129 56 64 2ms
XX.XX.21.129 56 64 3ms
XX.XX.21.129 56 64 3ms
XX.XX.21.129 56 64 19ms
XX.XX.21.129 56 64 12ms
XX.XX.21.129 56 64 10ms
XX.XX.21.129 56 64 12ms
XX.XX.21.129 56 64 5ms
XX.XX.21.129 56 64 3ms
XX.XX.21.129 56 64 5ms
XX.XX.21.129 56 64 9ms
sent=20 received=20 packet-loss=0% min-rtt=2ms avg-rtt=6ms max-rtt=19ms
We’ve open a ticket with Mikrotik ( [Ticket#2013050466000354] NV2 Latency problem )
We are waiting for useful answers
I’m also investigating how NV2 affects latency but I’ve found a post in this forum that seems to cover well this topic. It seems due TDMA logic, inactivy or low traffic nodes will face higher latencies but an active node will be properly time schedule and ICMP ping will demonstrate reasonable/true latency (lower than the inactive node).
In my example above we can see that the only ICMP is unable to obtain a good portion of time.
With 200Kbps bandwidth in transit latencies become a third
Other vendors do not have this problem also working in TDMA
Mikrotik support answer our questions
Ok, but can you improve this behaviuor in future release?
I see other vendor using tdma protocol with latency always stable to
1-2ms also with no traffic
with:
Hello,
we will check what we could do but it will not be so soon.
Regards,
Uldis