Skip to main content

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [List Home]
Re: [ecf-dev] ECF Helios plan

Hi Philipp,

Konradi, Philipp wrote:
Hi all,

here are some ideas on useful enhancements:

a) Extended JMS support
Concretely it would be:
- allow to configure JMS options like persistence, time-to-live, etc. on
a service-level.
- extend JMSContainerManager to manage multiple topics at the same time.
- enable usage of queues for Request-Reply communication between OSGi
services and clients

I agree these three additions would be good improvements...and I think will be fairly easy to implement...with only minimal API additions.

BTW, the last one (usage of queues), is now in place with the following container class(es): AbstractLBQueueProducerContainer and AbstractJMSQueueConsumerContainer (and appropriate JMS-impl subclasses...i.e. ActiveMQJMSQueueProducerContainer and ActiveMQJMSQueueConsumerContainer). This code is quite new (Aug 2009), but it's rather interesting, I think. I was planning on discussing/presenting this work as part of my talk at ESE.

b) Extend D-EventAdmin capabilities
Concretely it would be:
- allow to differentiate between events for local and/or remote
distribution (current impl tries to distributed any event so we got a
lot of NotSerializableExceptions for standard framework events)
- optimize filtering: filter already in JMS broker -> prevent
unnecessary communication

Yes, I agree these would be useful as well.

c) Improve D-OSGi / Remote Services Concretely it would be:
- enable declarative configuration of service exposure / consumption
(service properties / configuration tell which container type, port,
topic name etc. to use)
- enable smart proxies (e.g fail-over proxies which transparently
discover and call another service instance if one failed; e.g.
load-balancing proxy which balances the invocations amongst available
service instances)

Again, I agree that these would make nice additions as well. WRT the smart proxies...I have done some work on this already (using the load balancing stuff described briefly above), but there is much more to do.

So I'll let others' speak about their own responses to these...and listen a bit.

Thanks much Philipp (and Siemens),


Back to the top