[time-nuts] Amazon AWS time sync service

Tim Shoppa tshoppa at gmail.com
Mon Apr 29 17:38:02 UTC 2019


Long-term there is no drift to measure between regions. With each region on
its own GPS receivers and atomic clocks, it seems much more likely that
anything you measure will be asymmetries in network delays between regions.

Note that like Google has since 2008, AWS "Time Sync" also smears out leap
seconds.

My reading is that Amazon does their smear for 24 hours before the leap
second, and Google does it for 10 hours before and 10 hours after.

Google leap second smear reference:
https://cloud.google.com/blog/products/gcp/making-every-leap-second-count-with-our-new-public-ntp-servers

Amazon leap second smear reference:
https://www.zdnet.com/article/amazon-find-its-own-way-around-leap-second-problem-by-skewing-time-instead/

Tim N3QE

On Mon, Apr 29, 2019 at 1:02 PM Vlad <time at patoka.org> wrote:

>
> Hello Folks !
>
> Some time ago Amazon AWS has introduced "Time Sync Service"
>
> https://aws.amazon.com/about-aws/whats-new/2017/11/introducing-the-amazon-time-sync-service/
>
> Looks like it is just some NTPD (based on chrony). I am wandering how to
> measure drift between of AWS regions (internal clocks on EC2 instances).
> And how accurate that NTPD itself. Any ideas ?
>
> Regards,
> Vlad
>
>
> _______________________________________________
> time-nuts mailing list -- time-nuts at lists.febo.com
> To unsubscribe, go to
> http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
> and follow the instructions there.
>



More information about the Time-nuts_lists.febo.com mailing list