[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [HTCondor-users] Can we visualize HTC?
- Date: Mon, 8 Jul 2013 11:40:54 -0700
- From: Miha Ahronovitz <mihaa@xxxxxxxxxxxx>
- Subject: Re: [HTCondor-users] Can we visualize HTC?
Lars, this is a fascinating reply and you open my eyes - and I hope the eyes of many people on this list - to Sparklines , Flame -Graphs and so on.
From my point of view, imagine I can visualize all existing power-user of HTCondor. Then I can visualize all non-power engineers and researchers who may benefit from HTC use. Using a combination of their ability, motivation and triggers to act see Foggs model:
Then we adapt Bosco to capture this non-power user segment. The analogy is:
Text => Triggers => Tweeter
Email=> Triggers => Facebook
and why not:
HTCondor =>Triggers => Bosco ?
Now these visualizations go well beyond my personal interest, and ability. But I always look at bright people with ideas to help and teach me.
PS: To illustrate what we mean here is a special version of Bosco Âfor R! project users. R! Group in LinkedIn has 20,000 people and they have 50% share of stats predictive analysis market (SaS and SPSS being the other Âplayers). Slides prepared Âby Derek Weitzel.
We have researchers on the OSG and at our local universities that want to run R on our clusters
They donât know:
- The command line
- Cluster schedulers (Condor, PBS, LSFâ)
We want to create a seamless environment for them to run on remote clusters
Short summary page: Â
Descriptive Blog Post ÂÂ
On Mon, Jul 8, 2013 at 9:35 AM, Lans Carstensen <lans.carstensen@xxxxxxxxxxxxxx>
Since no one else bit on this....
If you're interested in visualization, study Tufte's work extensively:
His work on operational dashboarding culminated in his theory and
practice on Sparklines:
But to really get good visualization, you need to know what good
questions your people are asking. ÂFor that you might want to turn to
all the good business school research around Lean Manufacturing,
management of cycle times and work in progress, etc. ÂOr better yet,
actually ask your people what they're trying to measure and manage to
and help them do so.
Another natural device for an end-user watching a particular
dependency graph execute is the Flame Graph:
...and their workflow-optimizing question basically always comes down
to "am I getting the level of concurrency I anticipated from my graph
and teh system it's executing in, and if not - why not?"
-- Lans Carstensen
On Thu, Jul 4, 2013 at 12:35 PM, Miha Ahronovitz <mihaa@xxxxxxxxxxxx
> I discovered this video about visualization of data, where "each pixel
> conveys information"
> Can we visualize HTC, starting perhaps with the membership to this list? We
> can also visualize OSG, XSEDE.
> The questions is How? asked in a user-list deeply involved in CERN and Higgs
> particle discovery.
> HTCondor-users mailing list
> To unsubscribe, send a message to htcondor-users-request@xxxxxxxxxxx with a
> subject: Unsubscribe
> You can also unsubscribe by visiting
> The archives can be found at:
HTCondor-users mailing list
To unsubscribe, send a message to htcondor-users-request@xxxxxxxxxxx with a
You can also unsubscribe by visiting
The archives can be found at: