I had to try my fiber connection. I get 0.6 - 0.8 ms within our town which is not that big, < 10 km from me.
To Stockholm, which is only 80 or so km away if you measure straight line, is 3.5 ms away.
Iv'e always been curious as to what adds so much latency and what I did read was that its the network equipment itself mostly. The inherent latency in the fiber should be on the order of much less than 1 ms as far as I understand. 3.33 us per km, and 3,33 * 104 (more realistic distance) is only 346 microseconds.
Its weird though, using mtr its very clear that its the "long" hop that most of the latency, or at least 2.8 ms. So is the figures for speed of light in fiber wrong, or is there something else that adds up to a lot of latency in fiber connections? Is it the optical components?
Signals in fiber do not travel at c. They travel at lightspeed in glass. Light speed in a material is c/(the refractive index of the material). For commercial glass fiber, 1.5 is a reasonable estimate for it's refractive index. So the actual time for 1 km is ~5 us.
Traceroute work by routers decrementing the ttl as the packet passes. If a router doesn’t decrement, or if your packet gets bundled into an mpls, the hops may be hidden.
Then you’ve got the buffers - you could feed 5x100M streams on 1G carriers into a 1G uplink port and have plenty of bandwidth, but if all the packets of those 100M streams arrive in 100ms, then have no packet for the next 900ms, your outbound traffic will have to either delay your packets upto 900ms, or drop them.
To Stockholm, which is only 80 or so km away if you measure straight line, is 3.5 ms away.
Iv'e always been curious as to what adds so much latency and what I did read was that its the network equipment itself mostly. The inherent latency in the fiber should be on the order of much less than 1 ms as far as I understand. 3.33 us per km, and 3,33 * 104 (more realistic distance) is only 346 microseconds.
Its weird though, using mtr its very clear that its the "long" hop that most of the latency, or at least 2.8 ms. So is the figures for speed of light in fiber wrong, or is there something else that adds up to a lot of latency in fiber connections? Is it the optical components?