[time-nuts] Re: What everyday uses are there for accurate clocks?

[Keenan Tims]

>
> I can’t see how knowing who posted what first on Facebook, Instagram etc
> affects the user experience, when differences in time are any less than a
> second. So I am not convinced that timenut-grade oscillators are needed for
> that purpose!  I don’t know if accurate timing is important for network
> security.


They don't use tight time synchronization because they care about event
ordering as presented to the users (and all these social networks have
basically abandoned chronological feeds anyway, for improved addiction
susceptibility of their products). What does matter in this context is that
the massively distributed database cluster agrees on the order of updates
to its data structures, lest different nodes see a different, possibly
invalid, state. Very precise time sync means that instead of using locking
(blocking writes while reads occur) or a time oracle, it's feasible to use
'wall clock commit time' to order updates, which really simplifies and
improves performance and scalability of these systems. Google has more on
this in their TrueTime documentation:
https://cloud.google.com/spanner/docs/true-time-external-consistency

In this context of cloud-scale stuff, justifying PTP for it might be a
stretch, but it is also useful to have enough precision that timestamps can
resolve order of events across the large distributed system for tracing
purposes. Initial packet arrives at router, 100us later the OS receives it
and routes it to the application socket, 5ms later a related call is made
out to a distributed DB, this packet is launched at t+5.1ms and so on.


On Tue, 29 Nov 2022 at 23:01, Dr. David Kirkby via time-nuts <
time-nuts at lists.febo.com> wrote:

> On Tue, 29 Nov 2022 at 13:15, Hugh Blemings <hugh at blemings.org> wrote:
>
> > Hi,
> >
> > I'm sure others can speak in more depth than I, but birtually all major
> > cloud providers and/or services that make use of large datacentres
> > (think Facebook, Instagram, Linkedin etc. if you want publicly aware
> > examples) - will run their own NTP servers or other similar time
> > standards to ensure their geographically dispersed systems have a shared
> > understanding of what the time is.
>
>
> >
> > This is crucial for things as mundane as consistent end user experience
> > (who posted first?) to ensuring network security.
> >
> > Cheers,
> > Hugh
>
>
> I can’t see how knowing who posted what first on Facebook, Instagram etc
> affects the user experience, when differences in time are any less than a
> second. So I am not convinced that timenut-grade oscillators are needed for
> that purpose!  I don’t know if accurate timing is important for network
> security.
>
> As far as I am aware, of the 7 base SI units (metre, second, kelvin,
> candela, ampere, kilogram and mole), the second is the only SI base unit
> where low uncertainty is important to a large number of people. The
> instances where it is important seem to be limited to
>
> * Navigation systems (eg GPS)
> * Telecommunications
> * Financial transactions.
>
> Sure, timing is important for radio astronomy as someone mentioned, but
> that doesn’t have any practical impact to many people.
>
> Dave
> --
> Dr. David Kirkby,
> Kirkby Microwave Ltd,
> drkirkby at kirkbymicrowave.co.uk
> https://www.kirkbymicrowave.co.uk/
> Telephone 01621-680100./ +44 1621 680100
>
> Registered in England & Wales, company number 08914892.
> Registered office:
> Stokes Hall Lodge, Burnham Rd, Althorne, Chelmsford, Essex, CM3 6DT, United
> Kingdom
> _______________________________________________
> time-nuts mailing list -- time-nuts at lists.febo.com
> To unsubscribe send an email to time-nuts-leave at lists.febo.com