Re: [GridSim-users] dynamic job arrivals ...

[ALexander v. d. K. <va...@ho...>]

Hi Gilberto,
=20
No problem, glad to help.
=20
The execution time of a job depends on the size of the job (Gridlet length)=
, and the characteristics of the resource executing the job.
The faster the resource, the smaller the execution time. And the smaller th=
e job, the smaller the execution time.
=20
Basically, in a space shared resource, you can calculate the execution time=
, just by dividing the length of the Gridlet/job by the speed of the Proces=
sing Element (PE) of the resource.
However, you might not now the waiting time at the resource. If a job arriv=
es at a resource, and no free PE is avaiable (all are busy), it is placed o=
n a waiting queue. With no other information available, you can't calculate=
 the waiting time. This is possible if you exactly know all the job lengths=
 and submission times, and there is no (random) background noise at the res=
ource.
=20
You can configure the resource to behave exactly as you want by modifying t=
he GridResource.java source file, and using the modified and compiled versi=
on to create the resources.
=20
Kind regards,
Alexander


Date: Mon, 28 Jan 2008 08:55:52 -0300From: gil...@gm...To: yang=
yu...@hp...; ag...@ds...; va...@ho...Subject: RE: [=
GridSim-users] dynamic job arrivals ...CC: gri...@li...urceforge.=
net
Thanks Alexander, for your attention.and i'm sorry to take your time to rea=
d my email.

On Mon, Jan 28, 2008 at 8:42 AM, ALexander van der Kuijl <vanderq@hotmail.c=
om> wrote:

Hi Gilberto,=20
I'm not sure if I understand your question. With conclusion time, do you me=
an the response time (time between submitting and receiving the job)?
Yes.=20


It depends on how you design/program GridSim.If you know which jobs have be=
en sent to which resource, you should be able to calculate the various prop=
erties of the job execution (queue time, execution time, cost, response tim=
e, etc.).For space shared resources, this is of course easier than for time=
 shared resources.For 1 user you can do this within the NetUser class. Else=
, a central entity which controls the submitting of jobs to the various res=
ources might be a suitable approach. Does this answer (part) of your questi=
on?=20
Yes, :) Thanks.. but i don't understand one thing. The "Execution time" mus=
t be provide by user or is there other method for calculate the "execution =
time" in any resource (heterogeneous) ?




Kind regards,Alexander
Thanks, Regards,Gilberto Cunha=20

=20


Date: Mon, 28 Jan 2008 08:26:36 -0300From: gil...@gm...To: vand=
er...@ho...; ya...@hp...; ag...@ds...


Subject: Re: [GridSim-users] dynamic job arrivals ...CC: gridsim-users@list=
s.sourceforge.netHi, Yangyu, Agustin e Alexander.Thanks, for your answers. =
I was finding a framework for simulate any heuristics based in makespan for=
 Grid. I found the GridSim and Alea (http://www.fi.muni.cz/~xklusac/alea/),=
 and i read that  the Alea builded on the top of GridSim alow the dynamic j=
ob arrivals and i didn't understand.I wanted developing heuristcs which use=
 the "conclusion time" predicted for any job in any resource (e.g. Heuristi=
c Min-min, Max-min, Sufferage). I have another question, in GridSim, it is =
possible predict the conclusion time for job, before initiate the same? Is =
there some example for similar algorithm ?my english is not good too , Mr. =
Yangyu.Thanks for Attetion,Regards,Gilberto Cunha FilhoLaborat=F3rio de Sis=
temas Distribu=EDdosUniversisade Federal do Maranh=E3oBrasil
On Mon, Jan 28, 2008 at 6:18 AM, ALexander van der Kuijl <vanderq@hotmail.c=
om> wrote:

Hi Gilberto, As Agustin already mentioned, this is possible.It depends on y=
our NetUser.You can modify the source (NetUser.java) in such a way that ref=
lects your needs.For example, you can program that after receiving a job, a=
 new job will be sent.Or that the jobs are sent on random intervals. Kind r=
egards,Alexander

> Date: Mon, 28 Jan 2008 09:56:15 +0100> From: ag...@ds...> To: gil=
ber...@gm...> CC: gri...@li...> Subject: Re=
: [GridSim-users] dynamic job arrivals ...=20


> > Hi Gilberto,> > > Gilberto Cunha escribi=F3:> >> > Hi! my name is Gilbe=
rto, and I 'm studyng the GridSim for simulate > > scheduling algorithms.> =
>> > I am sorry to take your time to read my email.> >> > I would like know=
 if GridSim support dynamic job arrivals?> >> > e.g while one job is finish=
ing...other job can be submited for the Grid,> Of course, you can do this.>=
 > Regards,> > Agustin> >> >> > Regards..> > -- > > Gilberto Cunha Filho> >=
 gil...@gm... <mailto:gil...@gm...>> > LSD - Labora=
t=F3rio de Sistemas Distrbuidos> > UFMA> > --------------------------------=
----------------------------------------> >> > ----------------------------=
---------------------------------------------> > This SF.net email is spons=
ored by: Microsoft> > Defy all challenges. Microsoft(R) Visual Studio 2008.=
> > http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/> > ------------=
------------------------------------------------------------> >> > ________=
_______________________________________> > Gridsim-users mailing list> > Gr=
ids...@li...> > https://lists.sourceforge.net/lists/li=
stinfo/gridsim-users> > > > -- > =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D> Agustin Caminero> PhD =
Student> Computing Systems Department> The University of Castilla La Mancha=
, Albacete. Spain.> Phone: +34 967 599 200 Ext. 2693. Fax: +34 967 599 343>=
 http://www.i3a.uclm.es/> =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D>=20

Climb to the top of the charts! Play the word scramble challenge with star =
power. Play now!-- Gilberto Cunha Fil...@gm...LSD - Labora=
t=F3rio de Sistemas DistrbuidosUFMA=20


Need to know the score, the latest news, or you need your Hotmail=AE-get yo=
ur "fix". Check it out.


-- Gilberto Cunha Fil...@gm...LSD - Laborat=F3rio de Siste=
mas DistrbuidosUFMA -- Gilberto Cunha Fil...@gm...LSD - La=
borat=F3rio de Sistemas DistrbuidosUFMA=20
_________________________________________________________________
Climb to the top of the charts!=A0Play the word scramble challenge with sta=
r power.
http://club.live.com/star_shuffle.aspx?icid=3Dstarshuffle_wlmailtextlink_ja=
n=