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

[David Froble]

Stephen Hoffman wrote:
> On 2016-06-18 22:57:50 +0000, Kerry Main said:
> 
>> You are missing the point.
> 
> Um, no.   I'm not.  I know exactly what approach you are following, I am 
> very familiar with DT as implemented on OpenVMS, and I've seen it fail 
> utterly for products and for the competitiveness of more than a few 
> organizations.   General requirements and expectations — not losing 
> transactions — can be provided in different ways.   Sometimes in very 
> different ways.   Speeds, feeds, features, tools and designs change.  
> Sometimes for worse.  Often for better.   Existing software and existing 
> designs and existing assumptions either adapt, or they get replaced.  
> Failure to consider and examine and adopt newer approaches is how 
> various products and various organizations fail.   This is where I 
> certainly hope that new file system VSI is looking at is way past 
> SSD-ready and pondering newer non-volatile mechanisms and not 
> compromising for HDD support, for instance.

I think this is a good point, and I have the same hopes.  I'd ask, why re-invent 
the wheel?  VMS already has ODS-2 (and that other thingy I don't use) for 
rotating rust.  Something new should be FOR something new.

>  Where clustering adopts 
> faster host-to-host and server-level replication designs — whether 
> that's akin to MC or Galaxy or RDMA or otherwise; host-based server 
> shadowing, etc.
> 
>