How important is the choice of peers? I'm thinking of having 4 Suns (2 per location) for internal "stratum 1" and 26 Suns (2 per location) for internal "stratum 2". (All 30 are already dist'd across the national WAN doing other things.)<br>
<br>
My hypothesis is make several peer groups each using 3-5 systems in different facilities with low-latency / low-hop Layer 3 connectivity. (We have *many* VPNs, so Layer 2 dwgs. are useless.) These peer groups would also be cross-peered. The point is I think this architecture will speed the convergence rate on the time "voting" -- ie, the estimation / selection will be self-organizing within mult. regions simultaneously, then organizing at super-regions, etc. This is versus just throwing 25 peers at each one....<br>
<br>
My internal "stratum 1"s will be GPS-connected with backup local PPS clocks (from Passports)<br>
<br>
Am I seriously deluded? ;^) TIA
<br>
My hypothesis is make several peer groups each using 3-5 systems in different facilities with low-latency / low-hop Layer 3 connectivity. (We have *many* VPNs, so Layer 2 dwgs. are useless.) These peer groups would also be cross-peered. The point is I think this architecture will speed the convergence rate on the time "voting" -- ie, the estimation / selection will be self-organizing within mult. regions simultaneously, then organizing at super-regions, etc. This is versus just throwing 25 peers at each one....<br>
<br>
My internal "stratum 1"s will be GPS-connected with backup local PPS clocks (from Passports)<br>
<br>
Am I seriously deluded? ;^) TIA