Ask writes: > My thoughts are aimed more towards server operators than that of NTP users. > Let me explain: If we publish the numbers in real time, it may encourage > more than the (current) 5.5% of server operators to spend the ~$100 and a > few inches of solder it takes to wire up, say a Garmin* GPS 18 LVC to their > existing > stratum 1 server. This would boost the number of clock "sources" > versus "repeaters" in the pool.
While it would increase the number of stratum 1 servers, it would in fact reduce the diversity of stratum-0 refclocks being used. Because we would have more identical GPS refclock sources. I would encourage a wider diversity of stratum-0 refclocks rather than a homogeneity of stratum-0 refclocks. There are many refclocks other than GPS and several are already represented in the pool. In my perfect world, the folks running stratum-2 boxes would choose their configuration such that they get their time from stratum-1 nodes with a purposefully chosen diversity of refclocks. In the past twenty years, the diversity of refclocks has diminished greatly. Comparing Mills' 97 survey with Minar's 99 survey there was a marked decrease in non-GPS refclocks; it has gotten even more unbalanced today, with my surveys giving >95% of surveyed and sane stratum 1's being GPS based. Tim. _______________________________________________ timekeepers mailing list [email protected] https://fortytwo.ch/mailman/cgi-bin/listinfo/timekeepers
