[Info-vax] Distributed Applications, Hashgraph, Automation

[Jan-Erik Soderholm]

Den 2018-02-25 kl. 11:39, skrev johnwallace4 at yahoo.co.uk:
> On Sunday, 25 February 2018 07:22:01 UTC, DaveFroble  wrote:
>> Craig A. Berry wrote:
>>> On 2/24/18 12:33 PM, Kerry Main wrote:
>>>
>>>> In OpenVMS, you could also build master LD containers to image backup to
>>>> new OS, customize and reboot. Maybe 15-30 minutes start to finish?
>>>
>>> Then another month of tinkering to figure out how to change the node
>>> name without breaking anything.
>>
>> This is the result of setting up some things once, and assuming they will not
>> change.
>>
>> One would hope VSI takes a critical look at such, and perhaps uses some database
>> to contain all such things.  Then there is still the question of whether such
>> data would be set at boot time, or could be changed on the fly.  That could get
>> sticky.  If not a re-boot, perhaps some other type of "refresh".  There would be
>> the question of other computers "knowing" the node name, and getting confused.
>>
>> -- 
>> David Froble                       Tel: 724-529-0450
>> Dave Froble Enterprises, Inc.      E-Mail: davef at tsoft-inc.com
>> DFE Ultralights, Inc.
>> 170 Grimplin Road
>> Vanderbilt, PA  15486
> 
> What are names used for (and useful for) in the context of
> computers and applications (and, if necessary, users)?
> 

You do need *some* reference to whatever you'd like to connect
to, I guess.

> How many of those uses are things that people outside the
> IT Department should care about?
> 
> How many of them are things that should be important to
> the innards of OS, rather than (say) some OS-independent
> distributed naming layer on top of the OS?
> 

Most used do not care which of the servers you are actually
"using" when you access www.google.com.

> Host names, for anyone outside the IT department, for
> example? In a seriously distributed environment, are
> host names as such not a rather dated and devalued
> concept? Perhaps they should even be deprecated (for
> things being designed from scratch)?
> 

And replaced with, what?

> Even back in the 1980s, in a terminal-centric environment,
> things like terminal servers allowed user-visible 'service
> names' to be distinct from IT-visible host names. A bit of
> 'terminal server magic' was all that was needed. For LAT
> users, or for telnet users (round robin DNS?).
> 

Very much like having an DNS A-record for the "host IP address"
and one or more DNS ALIAS-record for the "service names" (pointing
to the A-reord). If you always use the name of an ALIAS-record, the
IP address of the A-reocrd can change with no changes for the user.

It is just a differnt name. A "LAT service name" could be seen as
a IP DNS ALIAS record.

> What's the 'modern' equivalent, where what is needed is not
> just users talking to application services but applications
> talking to other applications, in a (semi?) standardised
> fashion? (The obvious legacy approach is to use well-known
> IPhostnames and wll-known IP portnumber/name but that's
> not really helpful for reasons that should be fairly obvious)
> 
> And why do the OS internals have to get involved in this,
> except to provide the necessary facilities in a suitably
> robust and trustworthy way?
> 
> As a historical side note, I'm thinking that back in the
> 1980s, there was a VAX VMS software product that did the
> *technical* stuff of changing the SCSnode and DECnet
> name and stuff like that as part of deploying what Kerry
> likes to call a 'golden image'. It might have been called
> VAX Remote Systems Manager or something like that, and it
> wasn't just intended for use within a VMScluster. No
> matter. Anyway, on top of that, there was still the licencing
> stuff, which DEC did one way, others did other ways (FlexLM,
> dongles, etc). Three decades later there still isn't a
> universally accepted licence management and enforcement
> mechanism.
> 
> DHCP and friends (mDNS etc?) may be part of a modern
> follow on. Or may not. But I'm struggling to see why
> a host name (as such) is still important (outside the
> IT department). Application service names? Different
> matter; they may well need to be meaningful, or at
> least pre-agreed.
> 
> To an extent, the same naming issue applies to storage
> (files etc). That data someone wanted, those files
> that need restoring, are they on C: or are they on
> banana$dka300:[john] or /usr/users/john, or what (and
> where)?
> 
> Enlightenment welcome.
>