My company provides physical layer optical transport equipment in the Metro space. Some emerging carriers provide SLAs based upon roundtrip latency from router at the POP to the router at the customer prem. Latency at the physical layer is dominated by the time of flight through the fiber. How does the latency at the physical layer manifest itself in the TCP/IP layers?
Tek-Tips is the largest IT community on the Internet today!
Members share and learn making Tek-Tips Forums the best source of peer-reviewed technical information on the Internet!
-
Congratulations SkipVought on being selected by the Tek-Tips community for having the most helpful posts in the forums last week. Way to Go!
How does latency at the physical layer relate to latency at the TCP/IP
- Thread starter 1light
- Start date
do want to know the efect on the aplication layer or the effect on the second and third layers?
you can write a book on the effect of latency.
a real quick Answer is the latency works out to be how fast data can be sent and recived but i get the feel you where looking for a little more. So long and thanks for all the fish.
you can write a book on the effect of latency.
a real quick Answer is the latency works out to be how fast data can be sent and recived but i get the feel you where looking for a little more. So long and thanks for all the fish.
- Thread starter
- #3
>... quantitative relationship between latency at the physical layer (dominated by the propagation of light down the fiber) and its effect on latency at layer 2/3.
Ok... so you're the layer 1 latency guy, eh? Well... I'm a NIC that uses an access method like CSMA/CD... I need to wait for the 'cable' to be free so I can transmit. If transmissions are really quick, I'll have a greater chance at accessing the network. Also... you can figure out the layer 2 or 3 one-way latency by simply calculating the bps rate onto the wire (don't forget to include any preamble and interpacket gap times when figuring in 'to the wire' latency). If we're able to do 1000 Mbps because we're over fiber, then we can divide by 8 to get the bytes/second and then figure on an Ethernet packet of 1518 bytes (then add in 8 for the preamble and another 12 bytes for the interpacket gap time).
Now do the same computation using only 100 Mbps and then 10 Mbps.... now you should see the actual number of maximum packets we can get on the wire. Hmmm... did I get off track a bit...?
Consider getting a network analyzer and looking at some roundtrip request/reply functions -- divide by 2 to get the approximate 1/way latency time. Also look over at NetDoppler at Cool latency/throughput tester...
Ok... so you're the layer 1 latency guy, eh? Well... I'm a NIC that uses an access method like CSMA/CD... I need to wait for the 'cable' to be free so I can transmit. If transmissions are really quick, I'll have a greater chance at accessing the network. Also... you can figure out the layer 2 or 3 one-way latency by simply calculating the bps rate onto the wire (don't forget to include any preamble and interpacket gap times when figuring in 'to the wire' latency). If we're able to do 1000 Mbps because we're over fiber, then we can divide by 8 to get the bytes/second and then figure on an Ethernet packet of 1518 bytes (then add in 8 for the preamble and another 12 bytes for the interpacket gap time).
Now do the same computation using only 100 Mbps and then 10 Mbps.... now you should see the actual number of maximum packets we can get on the wire. Hmmm... did I get off track a bit...?
Consider getting a network analyzer and looking at some roundtrip request/reply functions -- divide by 2 to get the approximate 1/way latency time. Also look over at NetDoppler at Cool latency/throughput tester...
- Status
Similar threads
- Locked
- Question
- Locked
- Question