SPB an Overview

This will be the first blogpost of a series of posts about SPB. We start with an general overview. At the moment there are big changes in the network industry. After using protocols like spanning tree for over a decade technology has evolved and fabric based solutions can do today all the stuff that we couldn’t do with spanning tree. Of course an Ethernet fabric is much more than just an upgrade of spanning tree. The two standards for Ethernet fabrics are TRILL and SPB.  I will try to get you an overview what SPB is and how it works. SPB stands for shortest path bridging and is the IEEE standard 802.1aq. There are two versions of SPB the SPBM and SPBV standard, at the moment it looks like that only SPBM is adopted by vendors and everything posted here will regarding to SPBM.
With SPB you can build large loopfree multipath equal cost topologys. That is achieved with the ISIS and the dijkstra algoryththm as controlplane in the backround.
Think of it like routing for MAC Adresses. Some of the behaviour that we know from L3 Routing can applied with SPB for MAC addresses on L2. So like we did in L3 routed networks you can use all physical layouts like star, ring, full mesh, squre or what ever you like. In this network all path will be used there will be no blocked or unused path. It is also impossible to have a loop in a SPB network. We also get with SPB some features that are a littlebit similar to what we have done in the path with MPLS. Packets that will be forwarded across your SPB fabric will become an extra “TAG” like the label in MPLS, when the packet is forwarded to its destination the “TAG” is removed on the access switch, so your SPB network will always look like one hop for connected hosts. The second benefit is that with the encapsulation process is added an service identifiyer or I-SID that brings us 16 million service IDs and eleminates the limit of 4000 VLAN IDs. On SPB you can run different services , so once SPB is running on a network you can configure the services you like to utilize. I will use here the Avaya terminologys for the services.

L2VSN: a virtulized layer 2 service over SPB, like a VLAN on conventional switches you can transport L2 connectivity over the SPB fabric. On the acces site VLANs will be mapped to an I-SID. The VLAN ID is only local significant here so you can map on different access switch a different VLAN IDs to an I-SID only the I-SID must be the same gloablly to provide L2 connectivity across multiple switches.

IPShortcut: a Layer 3 service over SPB using ISIS for routing for the Global Routing Table GRT

L3VSN: a virtulized Layer 3 service over SPB. Think of it like the vrf feature in MPLS. Multiple L3 services in diffrent vrfs.

Multicast: a virtulized L3 multicast routing service. In the past you needed addtional protocols like e.g. PIM for multicast routing. SPB has multicast routing onboard so you just need to enable it and have multicast routing over SPB.

T-UNI: the T-UNI is a subcategory of the L2VSN, wich I call the virtual cable. Basicly it transports everything as long it is ethernet from one given access port to another across a SPB facric.

My opinion on SPB: 

SPB is a straight forward protocol wich is easy to deploy and works extremly reliable. I have to admit that I was very sceptic (I am always sceptic when it comes to brand new stuff that nobody has tried before) when I started to investigate SPB in 2013. I really got excited about SPB after a 2 days workshop. After only 2 days of testing SPB in the lab I felt comfortable with the technology. Once SPB was rolled out on the lab equipment I tried everything I could thought of to break it, but SPB has not failed a single time. Most of the simulated failures where handled and converged by SPB so fast that it was hard to measure them. Based on my expirience I can also say that with SPB you have to spent less time to configure your network.SPB_PP

When you got interested in SPB you can also listen to the Packet Pushers Podcast Episode 210 SPB Implemtation Fundamentals, where I had a nice discussion with Rikki Cook, Ethan Banks and Greg Ferro about SPB.

 

http://packetpushers.net/show-210-spb-implementation-fundamentals/

About Dominik

Network problem solver
This entry was posted in All, Avaya, Blog and tagged . Bookmark the permalink.

7 Responses to SPB an Overview

  1. Glenn Flint says:

    Hi Dominic,

    Congratulations on your new blog 🙂

    Just to mention that SPB was originally derived from Cabletron’s Securefast Vlans which was introduced in the mid 90’s. An RFC was also published but It is a shame it has taken 15 years for something so similar to appear? John Roese worked at Cabletron and helped create Securefast Vlans and when he ended up at Nortel as CTO he introduced SPB.

    Sorry about the history lesson 🙂

    Good luck,

    Glenn..

    • Dominik says:

      Hi Glenn,
      year unbelivabel how long it takes sometimes from an idea to an actually working product. Like RFC 1925 Rule 11 says “Every old idea will be proposed again with a different name and
      a different presentation, regardless of whether it works.” This stayment is a decade old and still true 🙂

  2. kemal says:

    Hi Dominic,
    Regarding your comment onL2VSN, If we have 801.1Q tagged ports connected to SPBM cloud at different sites, let us assume
    site 1 (VLAN ID: 100,200,300)
    site 2 (VLAN ID: 500,501,503)
    site 3 (VLAN ID 600,601,602).
    Is it possible mapping different vlan id at one site to a different vlan id on another site by using same I-SID value (say I-SID:1001) by using AVAYA L2 switched VSN.?

    Thanks

    • Dominik says:

      Yes of course you can do that. The VLAn IDs are only local significant, everything that you map
      into the same i-sid will form up a L2VSN like one big L2 network.
      For your example it would look like this:
      site1
      i-sid 11111 vlan 100

      site2
      i-sid 11111 vlan 500

      site3
      i-sid 11111 vlan 600

      Now the VLANs 100, 500 and 600 that are on 3 different sites would look like one big L2 network.

      Cheers

  3. kemal says:

    Thanks for your quick reply. Sure this scenario is possible, but what if one trunk’s more than one vlans must be assigned to other site’s ports! vlans. Let me give an example:

    vlans 100, 500,600 must be first l2vsn
    vlans 200,501,601 must be second l2vsn
    vlans 300, 503,602 must be third l2vsn

    site 1
    port 1 vlan 100, vlan 200, vlan 300 maps to i-sid 11111

    site 2
    port 1 vlan 500, vlan 501, vlan 503 maps to i-sid 11111

    site 3
    port 1 vlan 600, vlan 601, vlan 602 maps to i-sid 11111

    Cheers

    • Dominik says:

      Not sure what you mean. On the trunks that connect your SPB switches wich each other there is nothing to configure.
      The Uplinks ports have only the SPB backbaone bvlans configured.
      On your local switches you do the VLAN to i-sid mapping. Everything that is mapped to the same i-sid ID will be form up
      a L2VSN wich means you have one L2 network.
      I don´t see the usaecase for mapping 3 VLANs on a local switch to the same same i-sid. I would use here on the local switch
      only one VLAN ID for all the clienst that have to be in the same L2 network.

  4. ollivier says:

    nice post

    could you advice me a book to understand thanks

Leave a Reply to ollivier Cancel reply

Your email address will not be published.

nineteen + 11 =

This site uses Akismet to reduce spam. Learn how your comment data is processed.