Can delay be measured on recieve? genius answers wanted!

binarysoulbinarysoul Member Posts: 993
An 'experienced' manager (?) really confused me today and I began to doubt my understanding of delay. Can someone please clarify this?

icon_evil.gif

My understanding is that delay is measured when data travels OUTBOUND and not INBOUND. In other words, if my PC sends data to server, we can meausre delay on the outbound, but once the response comes back from the server, we can't measure the delay. It doesn't make any sense to me if this is incorect.

Bottom line, if you get this in an interview, what's your answer?

******** Is delay measured on INBOUND, OUTBOUND or both? Why? *******

Comments

  • KalabinKalabin Member Posts: 64 ■■□□□□□□□□
    If you are talking about delay as in latency, then I would thing of it much like a ping through icmp. The source send's out an echo request to the host computer, the host then send's a echo reply back to the source which measure the entire time the trip took. So, I would see Delay as the total time taken to go over a given network. Please do correct me if im wrong.
  • JDMurrayJDMurray Admin Posts: 13,078 Admin
    Latency in electronics is typically measured as one-way latency. The period of latency is the time between initialing an action and the action occurring, such as flipping a light switch and the light bulb fully illuminating. Two-way latency is a common measurement in data communications, where call-and-response protocols, such as ICMP ECHO, are common.
  • binarysoulbinarysoul Member Posts: 993
    Thanks guys!

    Yes, I'm talking about delay as in latency, but on a WAN. JD makes a good point about one-way vs. two-way delay.

    Let's say you install a network monitoring software on the 'source network' and the total delay for a network transaction to/from WAN is 10 seconds. How can you measure delay of incoming traffic from WAN?
  • dtlokeedtlokee Member Posts: 2,378 ■■■■□□□□□□
    You can monitor delay by using SLA monitoring tools, typically specifing a packet size, and protocol you want it to emulate. But in general delay is made up of a number of different factors.

    Serialization Delay - this can be calculated by (bits sent)/(speed of the link bits per second) so for a 1500 byte packet over a T-1 you get (1500 * icon_cool.gif / 1544000 = .0077 (7.7ms)

    Propigation delay - this is the time it takes for the signal to cross the media (based off the speed of light) so : (length of the link in meters) / (2.1 x 10**icon_cool.gif so a 3000Km link would be (3000000) / 210000000 = .014 (14 ms)

    Queuing delay - this is a variable value but it refers to how long a frame is waiting in an output queue on an interface.

    Processing delay - again variable delay and refers to how long it takes for a device like a router to move a packet from an ingress interface to the egress interface.
    The only easy day was yesterday!
  • JDMurrayJDMurray Admin Posts: 13,078 Admin
    binarysoul wrote:
    How can you measure delay of incoming traffic from WAN?
    You would need to know when a packet left its transmission node and arrived at the receiving node. There would need to be a monitor at both ends of the transmission, each synchronized to the same precise time source. The logs of the two monitors would be compared and the delay time for each logged packet calculated.
Sign In or Register to comment.