Options

WRED on Core Routers: Why?

adam-badam-b Member Posts: 36 ■■□□□□□□□□
Hello All,

Why is the Core layer the place in which WRED is most commonly used? Would it not make as much sense to implement this at the distribution layer as well? I understand that part of WREDs job is to prevent global synch yet shouldn't its other benifits be taken advantage of at the distribution layer? Thanks for your reply!

Adam

Comments

  • Options
    Paul BozPaul Boz Member Posts: 2,620 ■■■■■■■■□□
    From the DocCD:
    WRED is useful on any output interface where you expect to have congestion. However, WRED is usually used in the core routers of a network, rather than the edge. Edge routers assign IP precedences to packets as they enter the network. WRED uses these precedences to determine how it treats different types of traffic.
    CCNP | CCIP | CCDP | CCNA, CCDA
    CCNA Security | GSEC |GCFW | GCIH | GCIA
    pbosworth@gmail.com
    http://twitter.com/paul_bosworth
    Blog: http://www.infosiege.net/
  • Options
    adam-badam-b Member Posts: 36 ■■□□□□□□□□
    Thanks for your reply. Though, its just as feasable to mark your DSCP on the incoming interface at your edge router and then utilize WRED on the output interface of that router pointing towards the core. I still do not understad the emphasis on using WRED predominatly on the core.
  • Options
    Paul BozPaul Boz Member Posts: 2,620 ■■■■■■■■□□
    The core is where most traffic passes, thus congestion is more likely to occur. The core is also more centralized than the distributed routers so better judgment can be made regarding the entire network's traffic patterns.

    [edit] If you're noticing tail drop on your distribution layer devices you may need to upgrade the links / hardware.
    CCNP | CCIP | CCDP | CCNA, CCDA
    CCNA Security | GSEC |GCFW | GCIH | GCIA
    pbosworth@gmail.com
    http://twitter.com/paul_bosworth
    Blog: http://www.infosiege.net/
  • Options
    Forsaken_GAForsaken_GA Member Posts: 4,024
    adam-b wrote: »
    Thanks for your reply. Though, its just as feasable to mark your DSCP on the incoming interface at your edge router and then utilize WRED on the output interface of that router pointing towards the core. I still do not understad the emphasis on using WRED predominatly on the core.

    Imagine this -

    You have a gig link leading into your network on the edge, but it never gets more than halfway used.

    You have 20 of those.

    Your core interlinks are 10 gigs

    Your edge routers will never kick in WRED, but your core links could very well become saturated, since they're they have to handle all the aggregate traffic, not just the incoming traffic from the edge.
Sign In or Register to comment.