[Info-vax] AWS introduces VMS PaaS

[IanD]

On Wednesday, July 12, 2017 at 9:39:03 AM UTC+10, Richard Maher wrote:
> On 11-Jul-17 10:13 PM, IanD wrote:
> > 
> > All done out of India too?  (sorry, I worked for an Indian IT outsouring company for more than 5 years. Their understanding of VMS was shall we say less than optimal - yes I realise the irony of that statement also points the finger at myself!)
> > 
> 
> I'll ignore the racial stereotyping ;-) For Oz Microsoft offers Sydney, 
> Melbourne, and Singapore as arbiter.
> 

I took the comment to be one of 'VMS in the cloud'. I was making specific reference to that combination and that a large chunk of AWS to my knowledge is supported out of India

Racial stereotyping? I didn't think so, just relaying my first hand experience of Indian outsourcing

Of the Indian IT outsourcing companies I have worked for and know people in (most of the big ones), they all struggle to find suitable VMS people for most roles and if your after RDB, you can almost forget it. 

I directly saw people being given VMS admin roles who had previously done a bit of VMS programming and zero administration. There were others hired for VMS support roles who had never logged onto a VMS box in their life!

How does this bode for AWS supporting VMS in the cloud on AWS was the point of my sarcasm

> > And VMS then heads down the path of yet another commodity platform
> 
> Another cash-flow string to its bow? YES!

Yes, it needs to offer what the rest is providing as a starting point, that is a given

But what specifically is VMS's unique strengths now?

What about VMS will turn heads and open wallets?

'General Purpose' may as well be 'General Porpoise' as far as the IT world is concerned

What specialty is VMS going to focus itself upon so as to make a platform from which to spring from? (and overtake the world!)

> > 
> > The question still remains, just what market segment is VMS trying to establish itself in?
> 
> What markets can it afford not to be in? "Workstation" is definitely 
> one, commodity server solution (x86 remember?) is NOT!
> > 

Workstation has somewhat given way to tablet / mobile except for power users and windows is still entrenched in the desktop space and I don't see that changing anytime soon

In time windows will be free, that's the trend. Enterprise may take a lot longer though

Trying to win market share in a space that is seeing the high volume players use their number advantage to grind down prices and have a cost advantage is going to be a tough tough nut to crack. 
Security is one such area that has VMS with a name advantage in this space (I didn't say an actual advantage for those wanting to put the boot into my comment)

Of course, VSI being small, has an advantage in terms of keeping overheads down, but the converse is it's hard to get economies of scale to channel money into things like future R&D etc if you are hard pressed on the sales side of the equation

Outsourcing surely must be part of the model going forward to assist here?

> > All we need now is a decent scripting language and we can move into the automation space also
> > 
> > What was that need for a VMS cluster in a cloud again?
> 
> A very elegant and feature rich yet failed technology :-( What was the 
> maximum nodes again? Geographic distances? There are other "better" 
> solutions.
> >

I hear ya! 

Clustering was bloody awesome in it's day

Cloud / linux 'clusters' continue to eat away at this advantage however

When one can spin up 200 servers in a few mins and fire off a large scale workload and then collapse it all when one's finished, you realise this is what clusters grew up to be while VMS was sleeping and dreaming it was the only real cluster

No one really cares if 10 of those 200 nodes went down while the workload ran, if it's on something like Hadoop, those dead nodes will have been covered by some other redundant nodes anyhow and if not, then the lost work portions would simply be picked up by the first nodes finished, so that one doesn't even have to stuff around with resubmitting something like a 'retained on error' batch job

I know we are still getting to first base with VMS on x86 and that will be the platform from which hopefully cheaper VMS options can spring from but people should not be under any disillusionment that x86 will somehow save VMS. x86 is only going to get VMS to the point of being able to start on the pathway of being competitive

Anyhow, I know with yourself, I'm just preaching to the converted about the need for VMS to keep moving forward and radically too!