[DotGNU]Re: Ditsributed server (was Re: Our blindspot)

[Chris Smith]

On Monday 11 February 2002 13:21, you wrote:
> Chris Smith <address@hidden> wrote:
> > We must make sure that there is a way of mapping multiple IP's to a
> > resource to handle multiple machines serving the same resource.
> > Also these IP's must be able to come and go.
> >
> > I've recently done some work where this was an issue and I've implemented
> > quite a nice system to solve the problems.  It's not perfect, and runs in
> > a very closed world, but works a treat.
>
> Sounds like you already have the first step towards a
> distributed server, or Distributed Execution Environment,
> implemented there.  Would you be willing to put this code
> under GPL and share it?

Well the world in which it runs is terribly proprietry.  The client is 
actually a perl module which provides access to 'middleware' resources for a 
bunch of CGI's (that are just doing client presentation stuff).

The distributed nature of the middleware resources comes from me building 
several server boxes and booting them on the same network.  It's a VERY 
manual job keeping them in sync - but that's totally acceptable in the 
environment in which they are deployed - and they don't change very often.
The sysadmin guy wrote a script that is invoked by the SNMP manager to modify 
the DNS server on the fly, and I took care of client IP address caching and 
the server SNMP trap generation.

TBH It'll be easier to re-implement the mechanism in the 'dotGNU and the way 
we see it' world, with whatever improvements/mods are deemed appropriate.
You do not want to use DNS.  Sure, the dotGNU Service Discovery Node (SDN) 
may hand out fully qualified domain names when asked for the location of a 
service, and that is presumably preferable to handing out raw IPs - but the 
SDN's are not DNS servers.  My scheme uses DNS as the 'thing that hands out 
host addresses' - dotGNU has an additional layer which adds the 'webservice' 
location/availablility info.

The client perl module (minus the proprietry stuff owned by Orchid 
Telematics) can be made available for sure - it atleast documents the IP 
table caching.  Doesn't do any feedback though.  That addition was added in 
view of what needs to be achieved with dotGNU.

The question is this.  Do we design our own Service Discover Node servers 
using protocols such as LDAP, UDDI etc or design our own.  If we use any sort 
of feedback then what we use depends on what facilities exist to give us the 
feedback support.

Does that make sense?

Chris

-- 
Chris Smith
  Technical Architect - netFluid Technology Limited.
  "Internet Technologies, Distributed Systems and Tuxedo Consultancy"
  E: address@hidden  W: http://www.nfluid.co.uk