08 the 20th-century-internet-peering-ecosystem

393 views

Published on

0 Comments
0 Likes
Statistics
Notes
  • Be the first to comment

  • Be the first to like this

No Downloads
Views
Total views
393
On SlideShare
0
From Embeds
0
Number of Embeds
2
Actions
Shares
0
Downloads
11
Comments
0
Likes
0
Embeds 0
No embeds

No notes for slide

08 the 20th-century-internet-peering-ecosystem

  1. 1. The  Structure  of  the  Yesterday’s  Public   Internet  The20 Century Internet thPeering Ecosystem ©2012  DrPeering  Interna>onal   Licensed  material  –  sales@DrPeering.net   hDp://DrPeering.net  
  2. 2. 3                                           THE  EARLY  INTERNET  PEERING   ECOSYSTEM   Provide  a  historical  perspec>ve  on  how  Internet  Peering  evolved               2  
  3. 3. Why  should  we  care  about  the  history  of  peering?  About  these  “peering  ecosystems?”   3  
  4. 4. My  Airplane  Story  •  NWA  Flight  DTW-­‐LAX  DelayàCancel  •  Rebook,  go  to  gate  34  (short  walk)  •  On  plane.  Boarding  disgruntles  •  Flight  aDendant:  “sit  anywhere”  •  Delayed…  Palpable  Anger    What  happened?   4  
  5. 5. What  happened?  •  Back  at  gate  •  Flight  was  Cancelled  •  “Room  for  15-­‐20  passengers”  •  “Proceed  in  orderly  fashion”  •  “Will  handle  as  many  as  we  can”  •  154  people  hauled  •  Line  à  special  case  override  à  Mob  Scene  •  Get  the  yellers  out  of  the  line  •  Aggression,  pushing,  shoving,  many  cops  called     5  
  6. 6. Who  is  responsible?  •  Airline  vs.  Passengers  •  Show  of  hands   –  ___  Airline  100%  Passengers  0%  responsible   –  ___  Airline  90%  Passengers  10%  responsible   –  ___  Airline  80%  Passengers  20%  responsible   –  ___  Airline  70%  Passengers  30%  responsible   –  :   6   Does  Context  drive  Behavior?  
  7. 7. The  Peering  Problema>c  •  Evolving  Internet  Peering  context   –  Posi>onal  power   –  Predictable  Behavior  •  This  is  a  talk  about  the  future  of  peering   –  Trajectory  from  the  past  •  From  a  Recent  Discussion  Paper   7  
  8. 8. History  of  Internet  Peering  Contexts   Access   Powers   Fat  Middle   • 2009à   • 2000-­‐2009   COMMERCIAL   • 1998-­‐2002   TRANSITION   • 1994-­‐1997   NSFNET   • 1987-­‐1994   ARPAnet   8  
  9. 9. 1st  Peering:  ARPANET  80’s  •  USENET/BITNET/X.25  could  not  connect  •  ARPAnet  limited  to  gov’t  &  contractors  •  CSNet-­‐NSF  project  to  connect  all  CS  depts  •  Spotlighted  AUP  problem  •  Bureaucra>c  complexity   –  SeDlement  of  financial,  admin,  contract  etc.  •  Peering  is  “interconnec>on  without  explicit   accoun>ng  or  seDlement”   Source:  hDp://www.interisle.net/sub/ISP%20Interconnec>on.pdf     hDp://drpeering.net/AskDrPeering/blog/ar>cles/Ask_DrPeering/Entries/2010/10/22_Origins_of_Internet_Peering.html       9  
  10. 10. Reference:  The  First  Peering  •  “This  disconnect  persisted  as  both  sides  assumed  that  any   agreement  to  exchange  traffic  would  necessarily  involve  the   seDlement  of  administra>ve,  financial,  contractual,  and  a  host   of  other  issues,  the  bureaucra>c  complexity  of  which  daunted   even  the  most  fervent  advocates  of  interconnec>on  –  un>l   CSNet  managers  came  up  with  the  idea  we  now  call  ‘peering’,   or  the  interconnec>on  without  explicit  accoun>ng  or   seDlement.”  Source:  hDp://interisle.net/sub/ISP  Interconnec>on.pdf   10  
  11. 11. NSFNET  –  ‘87-­‐’94   GrowthàTime  to  priva>ze  The  Internet    Core   Coopera>ve  Agreement  (Merit,IBM,MCI,…)   Open  Regional  Techs  Mee>ngs   NSF  sponsored  aDachments   Announce/celebrate  new  aDachments   11  
  12. 12. NSFNET  Transi>on  –  ’94-­‐’96   NSFNET  TransiBon  Model   Network  Service  Providers  (NSPs)   Network  Access  Points  (NAPs)   Rou>ng  Arbiter  (RA)     Route  Servers  (RSes)     Na>onal  Science  Founda>on  (NSF)   Would  fund  regionals  to  connect     to  NSPs  that  connected  to  NAPs   PacBell  NAP   AADS  NAP   MAE-­‐East*   Sprint  NAP  Strong  NANOG  Chair  model   Interconnect  became  “a  private  ma>er”   Chair  &  commercial  interests   12  
  13. 13. Commercial  Internet  –  ’96-­‐’98   NAPs=“Conges>on  Points”  Resale  &  Growth   LiDle  visibility/sharing   àPrivate  Peering   13  
  14. 14. Growing  Pains  Commercial  Internet  –  ’98-­‐…   NAPs=“Conges>on  Points”  Resale  &  Growth   LiDle  visibility/sharing   …Private  Peering   14  
  15. 15. So  the  big  guys  privately  peered  with  each  other   Point-­‐to-­‐Point  circuits  in  each  major  metropolitan  area   15  
  16. 16. Challenges  What  mo>vated  the  large  ISPs  to  move  out  of  the  NAPs?  A)  Conges>on  problems  at  the  NAPs  B)  NAPs  were  operated  by  compe>tors  C)  The  Na>onal  Science  Founda>on  (NSF)  carried  liDle  weight  in  the  commercial  Internet  D)  all  of  the  above   16  
  17. 17. This  led  to  the  Global  Internet  Peering  Ecosystem  •  Players  independently  act  in  their  own   interests  •  No  U.S.  Government  operated  backbone  •  Not  U.S.  Government  controlled  Internet   17  

×