Exclusive Servers running Microsoft’s cloud will be briefly out of sync with each other and official time standards on June 30, as they implement the leap second.
Microsoft has determined that clocks on tens of thousands of servers globally running Azure should switch to the leap second at midnight in the time zone where they are based.
A Microsoft spokesperson told The Reg: “Each individual clock will automatically adjust to the local clock rate to converge the entire platform to the correct time. Thus, for a short period of time, the Azure infrastructure components may have single second discrepancies."
This will be automatically corrected over time, the spokesperson reassured us, adding the firm doesn’t foresee availability or reliability problems hitting Azure.
Azure negotiated the last leap second in June 30, 2012, but was brought crashing down by a date bug that saw it unable to cope with an extra day, February 29, earlier that same year. The bug meant Azure went dark for 12 hours.
The world is divided into 24 time zones – most 60 minutes apart, but some 45 and a few with 30 minutes’ difference.
Microsoft’s time-zone-by-time-zone approach to implementing the leap second differs to the strategy being taken by two of the company’s biggest cloud rivals – Amazon and Google.
Both will smear the second across their systems’ clocks for a 24-hour period either side of the new second’s addition to Universal Coordinated Time (UTC) at midnight on June 30.
The idea is to ensure massive, multi-node distributed systems that run their clouds continue to operate without a hitch.
A difference of a single second could cause discrepancies in synching operations and in systems of record, like databases, where different systems claim to have the correct time.
Microsoft is relying on the Windows Time service – a system synchronised with UTC but not built to include the value of a leap second – used to keep time in Windows.
After a leap second occurs in Microsoft’s world, a client running Windows Time service is one second faster than UTC until the next synchronisation with UTC.
Windows Time service is a set of algorithms, which syncs computers in a given domain with an authoritative computer. Unix and Linux servers, like those at running AWS and Google, use Network Time Protocol (NTP).
Virtual machines running on Google Compute Engine will get the smear, which see all servers’ systems clocks slowed by about 14 parts per million, with the leap second added at the end of the smear window hopefully putting Google cloud back on civil time.
Incredibly, at Amazon, each second on its AWS servers will be slightly longer than the accepted definition of the standard second for the 24 hours, meaning AWS will exist in its own time zone. Servers will also run for up to half a second behind UTC for 12 hours either side of midnight on June 30. ®