nanog mailing list archives

Re: $1.5 billion: The cost of cutting London-Tokyo latency by 60ms


From: Valdis.Kletnieks () vt edu
Date: Mon, 26 Mar 2012 08:57:15 -0400

On Mon, 26 Mar 2012 12:16:53 +0200, Tei said:

I imagine a easier solution.  Use a random number generator in both
sides, with the same seed.  Then use a slower way to send "packets
re-sync" that will contain the delta from the generated number, to the
real actual number.

Congrats. You've just re-invented the crytpo method called "xor with a
pseudorandom bitstream".  And no, it doesn't minimize your round-trip
latency at all.

I suppose this speeds are needed for some "fast speed transaction",
that are leeching money from the background noise on the market.

Unfortunately, you are correct on that point.

Attachment: _bin
Description:


Current thread: