[Oz] Mozart or Java: simple fault tolerant distributed application

Seif Haridi seif at ps.uni-sb.de
Thu Jul 22 19:13:12 CEST 1999


Mathias,
I would like to understand better your problem:

Interpretation 1:
Do you have N coordination centers that send information, from time to time,
to ONE central server ?
The sever is then responsible of putting a total picture of the state.
The connection from the server to the centers as well as the centers sites
goes down and up.

Interpretation 2:
Do You have N coordination centers each is connected to all the others?
Information is broadcast from
each coordination center to all the others. Each center collect all
information and can visualize the whole state.
The connection among centers as well as the centers sites goes down and up.

Is any one of the above is correct interpetation of the problem or it is
something else?

    Seif

By the way Mozart supports fault-detection and handling both for site crashes
and network partitioning
as well as machanisms for handling faults based either on exception handling,
or by independent fault handing
based on watchers. This is in the distribution tutorial.
We will help you if we know more the structure of the application.

Mathias Picker wrote:

> Dear Oz-Users,
>
> I might face the task of realising a real simple hotline-like tool
> (_really_ simple), which will link 100 or so coordination centers.
> Basic idea is to log informations about incidents (many informatins per
> incident, entered at geographically distributed sites)
>
> One allmost-requirement :-) would be fault-tolerant distribution:
> one would expect that the network communication between sites as well
> as whole sites would fail rather more often then usual.
>
> If I wanted sites / groups of sites which are seperated by network
> failure to still add information to incidents (no modifications, only
> adding), add new incidents, and, on regaining network connectivity, to
> synchronize with all other sites (or at least the ones it regained
> connectivity to), so that I have again one unified image of what
> happened around the world:
>
> - what feature of Oz3 would make it a better choice then java?
>
> - how would I design such a beast in mozart?
>
>   (I'm thinking about one transaction object at each site, and a simple
>   synchronisation protocoll, but that's not really mozart-specific. Any
>   idea about a better design would be welcome)
>
> >From what I read, all the distribution features expect sites to be
> rather stable. So what feature of mozart might help me to do this _fast_
> (I'm talking about 8 weeks _max_; but the app itself is super-simple,
> maybe two to three weeks)
>
> Any ideas would be welcome.
>
> Thanks, Mathias
>
> P.S.:
> An alle deutsch-sprechenden: wie kann mann "Lagezentrum" ins Englische
> übersetzen, wenn es nicht "warroom" heissen darf?
>
> For the english speaking crowd: what would be a more civil word for
> "warroom"? I know thats the civil word, too; but it can't be used in
> this project for political reasons.
>
> --
>
>  Mathias Picker
>
>                             Mathias.Picker at virtual-earth.de
>                             Fon +49 172 / 89 19 381
>                             Fax +49 89  / 6973 6212
>
>   ------------------------------------------------------------------------
>    Part 1.2Type: APPLICATION/pgp-signature

-
Please send submissions to users at mozart-oz.org
and administriva mail to users-request at mozart-oz.org.
The Mozart Oz web site is at http://www.mozart-oz.org/.
-
Please send submissions to users at mozart-oz.org
and administriva mail to users-request at mozart-oz.org.
The Mozart Oz web site is at http://www.mozart-oz.org/.





More information about the mozart-users mailing list