[Srcomp] Comments on REQ-7-27-COMPRESS

[Ron Bonica <[e-mail]>]

Folks,

The following is a better statement Rationale:


  *   Large encapsulation headers consume excessive bandwidth
  *   Large encapsulation headers header impede many ASIC's ability to process packets at high data rates

The following is a better statement of the metric:

Encapsulation header sizes will be calculated and recorded in a table.  The table will contain one column for each candidate compression mechanism and one row for each of 12 use-cases.

In each use-case, the encapsulation header represents an SR path. In the first use-case, the SR path contains one segment. In each subsequent use-case, the number of segments increases by one. In all use-cases:


  *   All instructions except for the last instruction are topological instructions. They alternate between loose and strict routing.
  *   The final instruction is a VPN service instruction (e.g., END-DT4)
  *   Locators alternate between ULA and GUA addresses.

So, in use-case 5, the SR path would be:


  *   Segment 1: Locator ULA, instruction END
  *   Segment 2: Locator GUA, instruction END.X
  *   Segment 3: Locator ULA, instruction END
  *   Segment 4: Locator GUA, instruction END.X
  *   Segment 5: Locator ULA, instruction END.DT4

                                                                                                              Ron






Juniper Business Use Only