[palladio-dev] [sdq-dev] SimuCom/PCM: Multiple resources of the same ActiveResourceType in the same resource container

Steffen Becker steffen.becker at uni-paderborn.de
Fri Sep 26 09:02:51 CEST 2014


Hi,


first of all this is not implementation specific but intentionally. As 
you see in your discussions, the case of having the same resource type 
twice is semantically unclear, hence, invalid so far.

If you want to model some cases like the ones you mentioned, I see also 
the option to use infrastructure interfaces. There you can specify you 
need read/write operations and then have infrastructure components which 
dispatch to HDD/SDD. However, overall this is also no long term solution.

Also note, that the processing rate of active resources is a stoex. In 
case your HW processes with different speeds you should use a StoEx 
which gives rates according to the distribution of request types.

Overall, this requires thorough thinking and at least a master thesis 
doing a proper analysis of use cases and modelling option including 
their semantics.

Cheers,
Steffen


Am 26.09.14 08:36, schrieb Christian Stier:
> Hi Jörg,
>
> there is a multi-core attribute where you can specify the number of
> cores per resource. However, that unfortunately doesn’t cut it for all
> cases. Multi-socket processors are often (for x86 processors at least)
> rolled out with the same type in the same node, so you can somewhat
> approximate how they would behave by modelling them as one active
> resource with 2x the cores. However, you can’t really apply the same
> method for HDDs (which you also model as Processing/Active Resources).
> There are regularly cases for servers where you have both a slow disk
> drive and a faster SSD. There is the work-around of specifying another
> ProcessingResourceType for SSDS but that also comes at the disadvantage
> of not being able deploy a component that accesses general HDDs to a
> node that has an SSD.
>
>>  so adding another resource of the same type mightimplicate some kind
> of (advanced) resource dispatching.
>
> I agree with you on this part – I am currently trying to think of the
> least bad solution to the problem.
>
> Best regards,
>
> Christian
>
> *Von:*Jörg Henß [mailto:henss at kit.edu]
> *Gesendet:* Freitag, 26. September 2014 07:57
> *An:* Christian Stier
> *Cc:* palladio-dev at ira.uni-karlsruhe.de
> *Betreff:* Re: [sdq-dev] [palladio-dev] SimuCom/PCM: Multiple resources
> of the same ActiveResourceType in the same resource container
>
> Hi Christian,
>
> I think this is just implementation specific. Nonetheless, there are
> some problems due to the fact that ResourceDemands are usually only
> specified resource type wise, so adding another resource of the same
> type might implicate some kind of (advanced) resource dispatching.
>
> If you just want to model a dual core processor or alike, I remember
> there was some kind of replication attribute?!
>
> best regards
>
> Joerg
>
> --
>
> ______________________________________________________________________
>
> Karlsruhe Institute of Technology (KIT)
> Faculty of Informatics
> Institute for Programme Structures and Data Organisation
> IPD Reussner - Software Design and Quality
>
> Dipl.-Inform. Jörg Henß
> Researcher
>
> Am Fasanengarten 5, Building 50.34, Room 325
> 76131 Karlsruhe, Germany
>
> Phone: +49 721 608-46796 (-45993, secr.)
> Fax: +49 721 608-45990
> Email: henss at kit.edu <mailto:henss at kit.edu>
> http://sdq.ipd.kit.edu/
>
> KIT - University of the State of Baden-Württemberg and National
> Laboratory of the Helmholtz Association
>
> Am 25.09.2014 um 08:51 schrieb Christian Stier:
>
>
>
> Hi all,
>
> SimuCom currently assumes that only one resource of the same
> ActiveResourceType is hosted in the same ResourceContainer. Is this a
> PCM-wide assumption or does this fall into the category of
> implementation-specific limitations?
>
> The 'culprit' can be found in the resources.xpt generation template
> ActiveResourceAdd, where the names for all processing units are
> generated based the type of the generated resource:
> «this.activeResourceType_ActiveResourceSpecification.id.javaVariableName()»_provInterfaces
> . This causes issues when it is expanded multiple times in the same
> ResourceContainer for the same ActiveResourceType.
>
> I wanted to ask whether this is something that was done for a good
> reason, or rather is something that I can fix in good conscience.
>
> Best regards,
> Christian
>
>
>
> .........................................................
> M. Sc. Christian Stier
> Wissenschaftlicher Mitarbeiter
> Software Engineering (SE)
>
> FZI Forschungszentrum Informatik
> Haid-und-Neu-Str. 10-14
> 76131 Karlsruhe, Germany
> Tel.: +49 721 9654-632
> Fax: +49 721 9654-633
>
> stier at fzi.de <mailto:stier at fzi.de>
> www.fzi.de <http://www.fzi.de>
>
> .........................................................
> FZI Forschungszentrum Informatik am Karlsruher Institut für Technologie
> Stiftung des bürgerlichen Rechts
> Stiftung Az: 14-0563.1 Regierungspräsidium Karlsruhe
> Vorstand: Dipl. Wi.-Ing. Michael Flor, Prof. Dr. Andreas Oberweis,
> Prof. Dr. Ralf Reussner, Prof. Dr. J. Marius Zöllner
> Vorsitzender des Kuratoriums: Ministerialdirigent Günther Leßnerkraus
>
> .........................................................
>
>
>
> _______________________________________________
> Palladio-Bench developer mailing list. News and discussions on the
> Palladio software architecture simulator and related tooling projects.
> palladio-dev at ira.uni-karlsruhe.de <mailto:palladio-dev at ira.uni-karlsruhe.de>
> https://lists.ira.uni-karlsruhe.de/mailman/listinfo/palladio-dev
>
> _______________________________________________
> sdq-dev mailing list
> Liste der Studenten und Mitarbeiter von SDQ fuer SDQ-bezogene
> Ankuendigungen.
> sdq-dev at ira.uni-karlsruhe.de <mailto:sdq-dev at ira.uni-karlsruhe.de>
> https://lists.ira.uni-karlsruhe.de/mailman/listinfo/sdq-dev
>

-- 
Jun.-Prof. Dr.-Ing. Steffen Becker

University of Paderborn                Phone:  (+49 5251) 60-3320
Heinz Nixdorf Institute                Fax:    (+49 5251) 60-3530
& Department of Computer Science       Office: ZM1.02-10
Software Engineering Research Group    E-Mail: steffen.becker at upb.de
Zukunftsmeile 1
33102 Paderborn
Germany



More information about the palladio-dev mailing list