Re: [mdnsext] Hierarchical (host) domain names in mDNS?

["Albrecht, Harald" <[e-mail]>]

> Von: [e-mail] [mailto:[e-mail]] Im
> Auftrag von Andrew Sullivan
> (It's actually not a DNS label, either, actually, but never mind that.)

As I'm curious to learn more on this topic ... why do you think that it's not a DNS label? At least, I thought it to be a label?

 
> It seems that the environment you need is much more amenable to traditinal
> DNS use.  That's probably not the answer you want, but I think it's likely true.

I don't want a particular answer, I want to know. We may very well decide to not use mDNS, so it is really about finding out where the design "limitations" of mDNS are.

But no, my environment is rather openly hostile to traditional DNS use. Putting such rather useless discussions about reliability aside (as I trust DNS to be operated reliably) the main issue are the "business processes" of how host names (or rather domain names) enter DNS: they are not established and it will probably take at least half a decade at best to slowly introduce these environments to DNS.

> This isn't too surprising.  You can't really have an unmanaged namespace that
> is also managed, which is sort of what you seem to want.

The point here is that there is no need for management: the "management" has already taken place when designing the automation as we are in fact mirroring the technological hierarchical names in DNS. We aldready have means in place that allow all kind of funny technological names to be used in a DNS-compatible manner, that is, in a manner compatible with LDH. We do acutually use DNS, but only on the large scale. What we need in these environments, however, is a small-scale and completely administration-free solution. mDNS is appealing here ... but my show stopper seems to be the "hierarchical names" thing.

> One thing you could do is use some sort of trickery to make your labels flat,
> like service1_machine1.local, service2_machine1.local,
> service1_machine2.local, and so on.  This is slightly hideous because there's
> no convenient way to split up the service name from the machine name (one
> of the distinct advantages of te DNS is this hierarchical name space).  But you
> could generate such labels programmatically, I think.

Yes, of course, that is an option. However, we ditched it as it wouldn't really integrate with the existing systems. People when would need to know which names to use in which context: oh, if it's "local" (well, ".local."), then I need to use the underlines, otherwise, I'll use the dots. Recipe for disaster.

> 
> Better, I suspect, would be to create a mechanism for registering names of
> the services and machines in a local registry.

As I indicated above, a local registry would still be a single point of failure. Adding redundancy to it is a big effort. And customers would not accept it just for local use. For large-scale use, they will accept, but then we're talking about full-blown DNS. mDNS is a peer system that in this aspect of its design is exactly what will perfectly fit into the environments we are dealing with. It's just that we need hierarchical names, maybe only two or three levels below ".local.", but we need it.

Putting this details aside, I really enjoy the replies and suggestions here!

With best regards,
Harald Albrecht

--- I have to add the following due to local law; let's hope that there are no other places with a law forbidding to add the following ---
Siemens AG
Industry Sector
Industry Automation Division
Industrial Automation Systems
I IA AS CTO DH 1
Gleiwitzer Str. 555
90475 Nürnberg, Deutschland
Tel: [phone]
Fax: [phone]
mailto:[e-mail]

Siemens Aktiengesellschaft: Vorsitzender des Aufsichtsrats: Gerhard Cromme; Vorstand: Peter Löscher, Vorsitzender; Roland Busch, Brigitte Ederer, Klaus Helmrich, Joe Kaeser, Barbara Kux, Hermann Requardt, Siegfried Russwurm, Peter Y. Solmssen, Michael Süß; Sitz der Gesellschaft: Berlin und München, Deutschland; Registergericht: Berlin Charlottenburg, HRB 12300, München, HRB 6684; WEEE-Reg.-Nr. DE 23691322