[
Date Prev][
Date Next][
Thread Prev][
Thread Next][
Date Index][
Thread Index]
[
List Home]
|
Re: [milo-dev] [Q] How to keep the specified sampling interval accurately on the server
|
Don't forget about garbage collection, either. While the new G1GC
system is a great improvement, it is hard to get GC pauses consistently
under 20ms, so your application will have to tolerate occasional jitter.
On 10/31/2017 07:57 AM, Kevin Herron wrote:
> SubscriptionModel is only meant to server as a naive example of how
> you might sample your underlying data source.
>
> I expect users to implement their own more sophisticated approach to
> sampling. Ideally, one that accurately tracks how long sampling took
> and can adjust its next execution, as you suggest, and also samples
> the data without resorting to calling read() on the namespace.
>
> At work, our underlying data sources are PLCs, so I've developed a
> "RequestExecutor" system that handles scheduling these requests to
> the PLC and accounts for the latency of each request as it
> re-schedules the next execution. I'm not able to open source this
> work right now, but it's not all that complicated to build on your
> own.
>
> On Mon, Oct 30, 2017 at 9:34 PM, Shigeru Ishida
> <ishida_shigeru@xxxxxxxxxxx> wrote:
>
>> Hi Kevin,
>>
>> Is there a way to keep the specified sampling interval accurately
>> when developing a server ? For example, when 100 (msec) is
>> specified, it took 110 (msec), so the next sampling is done at 90
>> (msec).
>>
>> Looking at the code of SubscriptionModel.java, it seems difficult
>> because it uses ScheduledExecutorService.
>>
>> Regards,
>>
>> --Shigeru _______________________________________________ milo-dev
>> mailing list milo-dev@xxxxxxxxxxx To change your delivery options,
>> retrieve your password, or unsubscribe from this list, visit
>> https://dev.eclipse.org/mailman/listinfo/milo-dev
>>
>
>
>
> _______________________________________________ milo-dev mailing
> list milo-dev@xxxxxxxxxxx To change your delivery options, retrieve
> your password, or unsubscribe from this list, visit
> https://dev.eclipse.org/mailman/listinfo/milo-dev
>
