TOTEM Questionnaire

   The TOTEM project is a three-years research project funded by the
   governement of the Walloon region in Belgium. Almost 20 man.years,
   equally divided among the three participating research groups, will be
   devoted to the development of an open source toolbox containing
   Traffic Engineering tools suitable for IP/MPLS based networks.

   We would appreciate if you could help us to identify the main
   requirements for such tools by filling the survey below. The
   information that you provide will be used by the researchers working
   on the toolbox. We will also write an anonymised document that
   summarizes the responses without revealing ISP names.

   (An on-line version of this form is available at
   http://totem.info.ucl.ac.be/te_form.html)


1. Your network

   1.1 Which traffic information do you use to engineer your network ?
       Which combination of monitoring and TE tools do you use ?


   1.2 What kind of tools are important to you (please specify if it's
       (M)andatory, (U)seful, use(L)ess) ?

       [ ] Off-line tool
       [ ] On-line tool
       [ ] Centralized TE tool (server)
       [ ] Decentralized/distributed TE tool (in routers, e.g. CSPF)
       [ ] Strategic TE tool (sort of quasi static optimisation)
       [ ] Tactical TE tool (reaction to some events in real-time)


   1.3 Do you know your traffic matrix ?

       ( ) as a POP-POP intradomain matrix
       ( ) as a prefix-prefix interdomain matrix
       ( ) as an AS-AS interdomain matrix
       ( ) others, please explain.

       How do you get this traffic matrix ?

       ( ) Network tomography from link traffic measurements ?
       ( ) Direct measurement ?


   1.4 Where do you observe congestion ?

       [ ] On internal links (backbone, backbone to POP, intra-POP)
       [ ] On links with customers
       [ ] On links with peers
       [ ] On links with providers

       Can you quantify ?


   1.5 What kind of SLAs do you provide ?
       What is your major concern in terms of SLAs ?
       What are the SLAs you would like to provide in the future ?


   1.6 Could you provide some data on your network
       (rough number of nodes and links per area/AS, number of areas/ASes) ?

       What is your company's name ?


2. IGP-based Intradomain Traffic Engineering

   2.1 How do you currently set your IGP metrics ?

       ( ) CISCO standard (1 / capacity)
       ( ) CISCO standard (1 / capacity) + local manual tuning
       ( ) Other static rule, please explain.
       ( ) Optimization based on a demand prediction


   2.2 At which rate would you accept to change your IGP metrics to adapt
       to the traffic ?

       ( ) Any dynamic solution that would improve the traffic
       ( ) Periodical change: once every ... hours / days / weeks
       ( ) As few changes as possible to keep the control of the routing


   2.3 Do you allow multiple shortest paths for a single
       origin-destination pair ?


3. MPLS-based Intradomain Traffic Engineering

   3.1 Which MPLS features are important to you (please specify if it's
       (M)andatory, (U)seful, use(L)ess) ?

       [ ] Traffic optimization, load balancing ?
       [ ] Protection and restoration ?
       [ ] QoS-based routing of LSPs ?
       [ ] Congestion control ?
       [ ] Load sharing (among several parallel LSPs) ?
       [ ] Fair allocation of unreserved bandwidth ?

       How would you use on-demand LSPs ? At which expected frequency ?

       How would you use preemption levels ? Would you use them
       for QoS purposes ?


   3.2 If (G)MPLS is used for protection and/or restoration, which of
       these techniques are important to you (please specify if it's
       (M)andatory, (U)seful, use(L)ess) ?

       ( ) Edge-to-edge ? Local backup ?
       ( ) Pre-established backup ? With (shared) resource provisioning ?
       ( ) Bypass tunnels ? Detours LSPs ?

       Would any of these be traffic-class dependent ?

       Should protection/restoration be SRLG aware ?

       Should protection/restoration be preemption level dependent ?

       Would you still advocate the use of other kinds of link protection
       (e.g. SONET) ? Which one and why ?


   3.3 Do you expect a need for inter-AS tunnels in the future ? If so,

       For which applications ?

       How often do you expect to establish those tunnels ?

       What constraints could apply to these tunnels (in terms of
       resources, protection, restoration, ...) ?


4. BGP-based Interdomain Traffic Engineering

   4.1 Do you engineer your interdomain traffic ?
       If so, what is the most important for you

       ( ) Outgoing traffic ?
       ( ) Incoming traffic ?
       ( ) Both are important


   4.2 If you control your incoming traffic, which technique(s) do you use ?

       [ ] More specific prefixes ?
       [ ] Selective announcements ?
       [ ] AS-Path prepending ?
       [ ] MED ?
       [ ] Communities ? If so, how ?
       [ ] Others, please explain.


   4.3 If you control your outgoing traffic, which technique(s) do you use ?

       [ ] Weight ?
       [ ] Local preferences ?
       [ ] Route optimization tool ?
       [ ] Others, please explain.


   4.4 What is for you the major drawback of BGP ?

       ( ) Slow convergence
       ( ) No multi-path routing
       ( ) No quality information for the routes
       ( ) Low control on the incoming/outgoing routes
       ( ) Lack of traffic differenciation
       ( ) Others, please explain.

       Could you elaborate on this ?


5. Miscellaneous

   5.1 Do you have or do you foresee other traffic engineering
       issues ? Please describe them here ?




       ------------------------------------------------------------

   Do you want to be kept informed about the results of the TOTEM project ?
   If so, what is your e-mail address ?


   Thank you for your collaboration !!


   (Please send this form to bqu@info.ucl.ac.be)