[Info-vax] Container files are set to grow in importance on the IT road ahead

[Stephen Hoffman]

On 2016-02-06 03:59:30 +0000, David Froble said:

> However, my main point was, this ain't anything new.
> 
> And, if 90% is the same, with maybe 10% different, then as things now 
> stand, using a master disk and changing 10% afterward is better than 
> doing 100% for each system.
> 
> "As things now stand ..."


Sure.  We can even write most of this in BASIC, too.   It's just tons 
and tons of glue code.

And with no way to keep applications from getting tangled, and no way 
to contain malicious or vulnerable applications.   Not without making 
some OpenVMS kernel modifications.

The difference here being that this container code and the 
jails/sandboxes have been written and debugged and supported and 
integrated in other platforms.

So we're not all maintaining shed-loads of glue code, nor are we 
depending on developers to maintain conventions to avoid conflicts.

But then I remember similar arguments from the folks that were 
programming assembler — high-level languages were wasteful — from the 
1970s and 1980s.

Languages get easier, frameworks get higher-level, the need for 
assembler or for as much BASIC is reduced, deployments get faster, 
capabilities get larger, etc.

Even in a single-host, single-application, classic OpenVMS application 
environment, having a way to keep a vulnerable server process from 
being easily extended into a full-blown breach would be valuable.

So would having an easy way to install and update applications 
(securely, easily, and without FTPing kits around), particularly given 
that most applications are gaining dependencies.

Master images or golden masters or LD disk containers don't give you 
any of that.



-- 
Pure Personal Opinion | HoffmanLabs LLC