Hi Frank,
Thanks for this prompt reply.
I'd be glad to test this, but unfortunately, git urls to sourceware are not
accessible from my end (https is pretty restricted).
Any chance you'd have an alternative git repo via http / a repo on github ?
As a last resort, if you can attach to this mail a git patch with the parent
sha1 that'd be great.
Sorry for the inconvenience,
AG
-----Original Message-----
From: Ken McDonell [mailto:kenj@xxxxxxxxxxxxxxxx]
Sent: 28 July 2015 06:35
To: Frank Ch. Eigler
Cc: pcp@xxxxxxxxxxx; Aurelien Gonnay
Subject: Re: [issue] pmwebd graphite api performance issue
G'day Frank.
On 28/07/15 10:26, Frank Ch. Eigler wrote:
> ..
> That could work too, though doing it naively could involve O(#metrics**2)
> iteration, which could itself be heavy. ...
The old computational complexity meets reality trap ... not all
operations have the same cost.
I'll wager that for any von Neumann machine with real values of #metrics
(and #instances), an O(#metrics**2) times for loop with simple test
(p->indom == indom_of_interest && p->done == 0) will beat the pants off
any O(#metrics) times loop with pmGetInDomArchive() + free()
> ... Doing it the cache way in pmwebd
> turned out to be easy:
OK, sounds good.
Cheers, Ken.
___________________________________
The integrity of this message cannot be guaranteed on the internet. Therefore
EXANE cannot be considered responsible for the contents.
If you are not the intended recipient of this message, please delete it and
notify the sender.
This message is provided for information purposes only and should not be
construed as a solicitation or offer to buy or sell any securities or related
financial instruments.
Although it may contain some elements from publications produced by Exane's
research department, this message is not research.
Please consult our web site for important disclaimers and disclosures
concerning Exane's research. (http://www.exane.com)
___________________________________
|