Options

Hop: Router or interface?

binarysoulbinarysoul Member Posts: 993
I always thought a hop meant the physical router, but now I have a feeling that when packets arrive at one interface and exit another one on a router that is regarded two hops.

Is this accurate?

Comments

  • Options
    dynamikdynamik Banned Posts: 12,312 ■■■■■■■■■□
    I believe a hop refers to a change in the network or subnet, which would require a router (or perhaps an L3 switch).

    subnet a <> router <> subnet b <> router <> subnet c

    Subnet c is two hops away from subnet a and one hop away from subnet b.

    The short entry in Wikipedia seems to confirm this.
    Wikipedia wrote:
    With routing a distance in terms of topology and of a length that may be not specified topographically, i.e. one hop is the step from one router to the next, on the path of a packet on any communications network (on the Internet often discovered with pings or traceroutes). The hop count then is the number of subsequent steps along the path from source to sink.

    http://en.wikipedia.org/wiki/Hop_%28telecommunications%29
  • Options
    RTmarcRTmarc Member Posts: 1,082 ■■■□□□□□□□
    I've always considered a hop to be any point at which it passes through an administered interface.

    Your computer 192.168.1.2 > f0/0 R1 192.168.1.1 > f0/1 R1 4.2.2.2 > f0/1 R2 6.3.3.3 > f0/2 R2 192.168.2.1 > Computer B 192.168.2.2

    Personally, I'd consider computer B four hops away even though there are two physical routers. I know what dynamik is saying regarding changing networks/subnets and I think it boils down to a matter of personal preference.
  • Options
    mikej412mikej412 Member Posts: 10,086 ■■■■■■■■■■
    RTmarc wrote:
    Personally, I'd consider computer B four hops away even though there are two physical routers. I know what dynamik is saying regarding changing networks/subnets and I think it boils down to a matter of personal preference.
    I don't think you have a "personal preference" when it comes to TCP/IP and how it handles TTL processing.

    Same with Distance Vector Routing protocols. If hop count becomes "personal preference" then you'd be introducing lots of asymmetric routing issues into lots of networks -- or at least any network you introduced your "modified TCP/IP protocol" into.
    :mike: Cisco Certifications -- Collect the Entire Set!
  • Options
    dynamikdynamik Banned Posts: 12,312 ■■■■■■■■■□
    >tracert google.com
    
    Tracing route to google.com [64.233.187.99]
    over a maximum of 30 hops:
    
      1     1 ms    <1 ms    <1 ms  192.168.0.1
      2     6 ms     7 ms     6 ms  [my public ip]
      3     8 ms     *       11 ms  68.85.165.101
      4     7 ms     8 ms     *     68.86.232.177
      5     6 ms     7 ms     *     68.87.174.17
     ...
     17    48 ms    50 ms    49 ms  4.78.208.114
     18    50 ms    52 ms    51 ms  72.14.236.15
     19    70 ms    53 ms    62 ms  216.239.49.222
     20     *       54 ms    51 ms  jc-in-f99.google.com [64.233.187.99]
    
    Trace complete.
    

    The trace route output on my machine makes it look like it's each hardware device, not interface. I wonder if it's a difference between Cisco and MS terminology. I'm curious to see what others have to say.
  • Options
    mikej412mikej412 Member Posts: 10,086 ■■■■■■■■■■
    dynamik wrote:
    The trace route output on my machine makes it look like it's each hardware device, not interface.
    Except if there is an MPLS network (or several) in the path configured to "hide the hops." And switches are hardware devices too, but they don't count unless they are configured as layer 3 switches (and the packet goes though a layer 3 configured interface).

    And then there are tunnels that can hide lots of hops between tunnel endpoints.
    :mike: Cisco Certifications -- Collect the Entire Set!
  • Options
    RTmarcRTmarc Member Posts: 1,082 ■■■□□□□□□□
    mikej412 wrote:
    RTmarc wrote:
    Personally, I'd consider computer B four hops away even though there are two physical routers. I know what dynamik is saying regarding changing networks/subnets and I think it boils down to a matter of personal preference.
    I don't think you have a "personal preference" when it comes to TCP/IP and how it handles TTL processing.

    Same with Distance Vector Routing protocols. If hop count becomes "personal preference" then you'd be introducing lots of asymmetric routing issues into lots of networks -- or at least any network you introduced your "modified TCP/IP protocol" into.
    I don't seem to have been making any commennts on how "TCP/IP and how it handles TTL processing." I was commenting on the terminology of the word "hops" and how I perceive it from within an Operating System.
  • Options
    networker050184networker050184 Mod Posts: 11,962 Mod
    What mikej is saying is there is no room for perception on hops, it is a standard.
    An expert is a man who has made all the mistakes which can be made.
Sign In or Register to comment.