[occi-wg] About clock synchronization in the cloud

[Augusto Ciuffoletti]

Thank you for your replies, first...

Here my point is that clock synchronization is irrelevant in many (possibly
all) current use cases, but is essential for others that are just potential.
The user that hits such a use case is unable to implement clock
synchronization in the cloud (as far as I know), so he/she may be happy to
pay an extra to have a clock within 1 second from UTC on certain VMs, and
add a new use case. The simplest idea that comes to my mind (maybe wrong) is
to provide clock specs through a mixin.

The boundary of "what is subject to a SLA" is subtle. But a feature included
in a SLA is somewhat static, and does not (necessarily) go through the user
interface. Clock synchronization is a switch you can set on/off, and that
requires specific actions on the side of the IaaS provider.

I would be happy to know if anybody experienced the case where a lack of
synchronization in the cloud was a problem. If only  I'd know what PDC is
...

Augusto

2011/9/20 Edmonds, AndrewX <andrewx.edmonds at intel.com>

> I think Alex hits this on the nail by noting "as part of the [...] SLA",
> which would be then related to the provisioning of the compute resources
> but
> not necessarily to the current OCCI spec where consideration for SLAs is
> not
> (yet?) present.
>
> Andy
>
> -----Original Message-----
> From: alexander.papaspyrou at tu-dortmund.de
> [mailto:alexander.papaspyrou at tu-dortmund.de]
> Sent: Tuesday, September 20, 2011 11:27 AM
> To: Edmonds, AndrewX
> Cc: augusto.ciuffoletti at gmail.com; occi-wg at ogf.org
> Subject: Re: [occi-wg] About clock synchronization in the cloud
>
> From my point of view, this is out of scope. Getting the hardware clock
> "right" is a job that has been solved on the operating system level, using
> Distributed Systems protocols like NTP.
>
> I agree that, for many services, exact time is crucial (try to deliver a
> PDC
> with a broken clock.), but this would be either something that
>
> a) may be part of an operating system template mechanism, where you know
> that such basic services are available or
> b) is ensured through quality of service parameters (as part of the
> contract
> or SLA between consumer and provider)
>
> Making it an attribute seems to be the better of the two options for OCCI,
> but it still doesn't make that much sense -- what does an attribute on the
> clock (which way ever this attribute is structured) tell me other than that
> the clock is there and in sync (with a drift of x and skew of y) at any
> time, and why would I want to discover this information via the OCCI
> interface?
>
> Just my two cents.
> -Alexander
>
> Am 19.09.2011 um 22:39 schrieb Edmonds, AndrewX:
>
> > Sounds like a good discussion topic! It might be a little out of scope
> for
> OCCI, but who knows where the discussion might lead to or what it might
> highlight.
> >
> > Andy
> >
> > From: occi-wg-bounces at ogf.org [mailto:occi-wg-bounces at ogf.org] On
> > Behalf Of Augusto Ciuffoletti
> > Sent: Monday, September 19, 2011 9:33 AM
> > To: occi-wg at ogf.org
> > Subject: [occi-wg] About clock synchronization in the cloud
> >
> > Dear all,
> >
> > I was wondering whether the features of the system clock of a virtual
> instance (compute or storage) may be of interest for the user. I wrote down
> 500 words on the topic (see Google doc below), and I'd be pleased to know
> your opinions, taking advantage of the intense "brain storm" during next
> OGF.
> >
> > https://docs.google.com/document/pub?id=1imafqyrVscdTbdfjngJ4Xfbg1DIoe
> > bdyzyQSPmDgOCY
> >
> > Bye
> >
> > Augusto
> >
> > -------------------------------------------------------------
> > Intel Ireland Limited (Branch)
> > Collinstown Industrial Park, Leixlip, County Kildare, Ireland
> > Registered Number: E902934
> >
> > This e-mail and any attachments may contain confidential material for
> > the sole use of the intended recipient(s). Any review or distribution
> > by others is strictly prohibited. If you are not the intended
> > recipient, please contact the sender and delete all copies.
> > _______________________________________________
> > occi-wg mailing list
> > occi-wg at ogf.org
> > http://www.ogf.org/mailman/listinfo/occi-wg
>
> --
> Alexander Papaspyrou
> alexander.papaspyrou at tu-dortmund.de
>
>
> -------------------------------------------------------------
> Intel Ireland Limited (Branch)
> Collinstown Industrial Park, Leixlip, County Kildare, Ireland
> Registered Number: E902934
>
> This e-mail and any attachments may contain confidential material for
> the sole use of the intended recipient(s). Any review or distribution
> by others is strictly prohibited. If you are not the intended
> recipient, please contact the sender and delete all copies.
>
>


-- 
Augusto Ciuffoletti
Via Milano 118
38122 - Trento
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.ogf.org/pipermail/occi-wg/attachments/20110920/585a6962/attachment.html