Hi Thomas,

Great initiative, which i fully support. 

I'd like to make two points, which is nothing new to the
feedback that you already received: 1) what TC and Andreas say
in different ways is true: there is not a reliable, featureful,
well supported frontend to pmacct. So in essence lots of room
there; 2) agree with Karl about reconfiguration API. Moreover
consider: all maps, which is supposedly the content prone to
change over time, are all already realoadable at runtime with
no loss (via a UNIX signal); what you can't do is reload the
base configuration, ie. instantiate a new plugin, change config
for an existing plugin, upgrades, etc: this part, i believe, is
much less prone to change once you put the setup in prouction.

Cheers,
Paolo

On Tue, Sep 16, 2014 at 09:25:19AM +0000, Thomas King wrote:
> Dear all,
> 
> we at DE-CIX think about adding new features to pmacct that are needed from 
> our side. As pmacct is open-source we would like to release our additions as 
> open-source as well. To assess if our features are of general interest I 
> would like to discuss our features with you. The questions for us is here: 
> Are the features we might add be supported by the community?
> 
> We are thinking about adding features in the following categories:
> - High availability: We would like to enhance pmacct in such a way that if 
> pmacct dies on one machine another will take over automatically. We assume 
> that existing open-source tools (e.g., heart-beat) can be combined with 
> pmacct in order to achieve this.
> - Reconfiguration via API: As we want to use pmacct in a dynamic environment 
> we want to be able to change the configuration via an API without restarting 
> pmacct.
> - Report and statistic engine: We would like to define a set of reports 
> (e.g., throughput in bits per seconds pro MAC address or interface) which can 
> be displayed on a website including fancy looking graphs. This will probably 
> not be included into pmacct directly instead it will be a standalone tool 
> that relies on pmacct. 
> - Notification: If a threshold is met we want to be informed by mail (e.g., 
> the throughput for an interface reaches a certain level we want to receive a 
> mail). This will probably not be included into pmacct directly instead it 
> will be a standalone tool that relies on pmacct.
> 
> I am aware that these feature descriptions are somewhat high-level. We 
> already created a more detailed technical description of what we need to add 
> to pmacct so that it fits our requirements. We will share this list later but 
> we do not want to steer the discussion to much in one direction.
> 
> Any feedback to our ideas is highly appreciated!
> 
> Best regards,
> Thomas
> 
> -- 
> Dr. Thomas King
> Manager Research & Development
> 
> DE-CIX Management GmbH | Lindleystraße 12 | 60314 Frankfurt am Main | Germany 
> | www.de-cix.net
> Phone +49 69 1730902 87 | Mobile +49 175 1161428 | Fax +49 69 4056 2716 | 
> [email protected]
> Geschaeftsfuehrer Harald A. Summa | Registergericht AG Koeln HRB 51135
> 



> _______________________________________________
> pmacct-discussion mailing list
> http://www.pmacct.net/#mailinglists


_______________________________________________
pmacct-discussion mailing list
http://www.pmacct.net/#mailinglists

Reply via email to