[Planetlab-users] Synchronizing clock and measuring
vivek at CS.Princeton.EDU
Tue Jul 5 17:38:11 EDT 2005
I'm not sure if Neil intended to imply this or not, but it should be
noted that the code in question isn't suitable for tightly synchronizing
two different nodes. It's really designed to provide fine-resolution
time information using the hardware cycle counter, without worrying
about NTP-based adjustments to the user-visible clock. Back in PlanetLab
v2, clock drift was a much bigger issue, and it wasn't uncommon to find
clocks moving backward during measurement periods. This code was
designed to eliminate this problem, but not to care about how close your
local clock was to the actual time.
Neil Spring wrote:
> On Jul 5, 2005, at 1:51 PM, Neil Spring wrote:
>> On Jun 27, 2005, at 4:09 PM, Jawwad Shamsi wrote:
>>> I am trying to synchronize clocks and measure one-way latency
>>> between end-points on planet-lab. Has anyone implemented related
>>> tool or service? Please let me know.
>> I don't know of any such service. If you're thinking of implementing
>> one, you might want to be familiar with Aki Nakao's NTP-independent
>> reimplementation of gettimeofday, described at:
> Ooops! wrong clipboard buffer...
> yeah, that's the right one.
> "when reporting a missing url, please do not describe your url as 'all
> urls look alike.'"
> Users mailing list: Users at lists.planet-lab.org
More information about the Users