[Info-vax] RMS record metadata, was: Re: Re; Spiralog, RMS Journaling (was Re: FREESPADRIFT)

[Stephen Hoffman]

On 2016-06-19 18:18:15 +0000, Kerry Main said:

>> Hoff: Isn't this the flip-side of that "blue ocean" stuff you're fond 
>> of pointing at?
> 
> Nope - the blue oceans is for new service offerings.

And doesn't a "blue ocean" have to put together existing and new tools 
in a way that allows those services to be delivered?

That involves both knowing what the folks want before they want it., 
and then getting there in a sustainably-profitable way.

That — in IT — usually means adapting some software and reusing other 
software — because nothing in IT is ever entirely "from scratch".   
It's always built on existing tools and processes.

If VSI wants to take a product and associated services — in a new 
direction — because the existing market for OpenVMS isn't exactly 
seeing substantial growth — they're going to have to adapt.

For your case, if Stark is going to have high I/O loads, then you're 
likely going to have to implement geographically-dispersed servers both 
for redundancy and particularly for lower latency, and then figure out 
how to coordinate and synchronize those.  This as clustering inevitably 
runs into latency.  At higher I/O loads, you're either buying piles of 
really fast storage hardware — thankfully for folks other than the 
major vendors, the prices on that are dropping — and quote possibly 
replicating your data in battery-protected servers at the server level, 
and not at the "traditional" disk spindle or array level.  That's 
before looking at contending with what will likely be greatly shifting 
I/O loads due to game releases and seasonal variations, and this gets 
interesting as hardware and network over-provisioning gets hugely 
expensive; that comes right off the bottom line.    If you do start 
toward replicating data centers, then you're going to be looking at 
DVCS and other and newer tools, at scaling deployments and other topics 
— most off which I've been pointing you at, too.

But then...   Blue Oceans ain't ever "all-new".   Nothing in IT is ever 
that way.   Never all-new.   It's all built on existing tools.  Bolted 
together from old tools, and usually from some of the best of new ideas 
and new tools, and with a dollop of insight.    As with most other 
projects, the idea behind "Blue Ocean" is worthless.   It's the 
execution and the timing that matters.   Then — for a successful 
endeavor — dragging those old tools and now-old ideas forward fast 
enough that some new competitor doesn't show up and "Blue Ocean" your 
business.

But enough of the who-moved-my-cheese.

Please: go learn about server-level replication, about DVCS tools, and 
about newer approaches toward deployments and provisioning.   The IT 
business moves forward quickly, after all.


-- 
Pure Personal Opinion | HoffmanLabs LLC