[Ace-users] [tao-users] ACE (and TAO) timers for Real-Time

[Mauricio A. Araya L�pez]

On Wed, Feb 27, 2008 at 02:28:48PM -0800, Thomas Lockhart wrote:
>>  An other little question, is if the real-time performance of the ACE
>>  Timer_Queue is ensured for distributed objects (if the proper network
>>  is setup).
>
> I would guess that this is not so much dependent on ACE as it is on your 
> definition of "real-time" and which network stack you use.

My definition of "real-time" is just to be predictable. I am testing how
far RTOS can goes for distributed real-time systems.

> Some would say that you would need something like RTDM within Xenomai or 
> RTAI to ensure network real-time performance, but perhaps your needs are 
> not quite that stringent.

Is for a master thesis, I am comparing platforms (Linux with RT patches,
QNX and VxWorks). The idea is to check for each platform how predictable
they are for distributed aplications. ACE and TAO offers me portability,
so that the reason why I am using the middleware.

> You might achieve "good enough" by using local timing services and local 
> NTP to get within a few microseconds, but of course that would not 
> guarantee execution ordering between distributed processes.

Yes, that is the other paradigm that I am exploring, and till now is
giving much better results. But the idea of using the RT-CORBA and
the ACE patterns is quite tempting also, as you must know.

Also, the framework that i am using (ACS), is based on TAO,
so i have no more options that use ACE. When I said "proper network",
this could be just unix sockets over a single host. TAO and ACE, 
is not only a CORBA/Network library to me, is a pattern factory
that could help to build complex real-time systems without leaving
a single host.

Bye!

-- 
Mauricio Araya L. 
ACS-UTFSM Team Leader, cel: 0-81564903
Est. Magister en Ciencias de la Informatica UTFSM, Valparaiso, Chile
counter.li.org: #249395, http://www.inf.utfsm.cl/~maray