[Info-vax] free shell accounts?

[Stephen Hoffman]

On 2015-01-23 12:47:41 +0000, Stan Radford said:


> I just didn't understand how the deathrow vax cluster could really be a 
> cluster if the machines presumably were separated across the world or 
> even one country between
> various hobbyist boxes without sharing disks. I guess that means it was 
> a cluster in the sense VMS itself was available but not that people 
> could develop and test their code across machine failures.

The Deathrow cluster is located in a rack, in Florida.

Deathrow did once have a VAX or two (one was named MANSON, IIRC), but 
that hardware was retired a while back.

OpenVMS clustering is supported for rather longer distances than a rack 
or a computer room, however.  When last I checked the associated 
Software Product Description (SPD), ~242 kilometers was the baseline, 
and it was fairly easy to get configurations to work at ~800 kilometers 
separation with a little help, and one not-entirely-stable cluster 
between servers in Nashua and Bengaluru was demonstrated.  HP is fond 
of pointing out that certain aspects of OpenVMS data protection have 
been demonstrated at distances of 90 thousand  kilometers, too — that 
using communications delay simulation, AFAIK / IIRC.

The particular configuration for Deathrow is not particularly robust 
(it's two nodes and no shared interconnect, to start with), but a more 
appropriate hardware configuration can survive machine failures and 
remain available.  This through clustering for server and data access 
and for application uptime, as well as using DNS-level and IP-level 
mechanisms for remote access.


-- 
Pure Personal Opinion | HoffmanLabs LLC