Skip to main content

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [List Home]
Re: [ecf-dev] OSGI Remote Services And Whiteboard Pattern

On 5/7/2014 9:48 AM, Wim Jongman wrote:
If you keep the fallacies of distributed computing in mind [1] then the "Whiteboard Pattern" (don't call me, I call you) works just fine for remote services.

IMHO the dynamic nature of remote services are almost per definition suitable for the "Whiteboard Pattern".

[1] http://en.wikipedia.org/wiki/Fallacies_of_Distributed_Computing

Sure...I certainly agree with Wim that the dynamic nature of OSGi remote services make it very well suited for building distributed systems...with the whiteboard pattern or not. I hope I haven't conveyed otherwise...as it wasn't at all my intention to do so.

But IMHO 'keeping the fallacies of distributed computing in mind' can be a very...sometimes overly...complex thing to do...especially when you are talking about run-time behaviors associated with distributed systems...e.g. problems associated with timing and reliability/partial failure (e.g. distributed consensus [1]).

To me, this means that even with the wonderful OSGi service registry, remote services...and the supported dynamics...there is/will always be room for additional...complementary IMHO...models....like pub/sub/group models...to help app/system builders deal with the problems in distributed systems. An example: weak failure detectors for dealing with partial failure. Like many things, I don't think this is a 'one model will fit all application needs' situation.

Scott

[1] http://en.wikipedia.org/wiki/Consensus_(computer_science)




Back to the top