[Info-vax] Estimates of long-term availabilty of 600MHz DS10s

[Paul Sture]

On Tue, 15 May 2012 07:09:54 -0700, abrsvc wrote:

> The quality of the service needs to be understood as well.  In my case,
> we were paying $2500 per month for hardware support.  With
> repair/replacement times of 4 hours that were never met because of the
> lack of Alpha knowledge, we dropped that service.  Spending 4 months
> worth of "hardware maintenance" money, I have enough replacement
> hardware (tested and verified) to replace our entire copllection of
> production systems with a few spare parts left over.  The cost savings
> along justified the purchases.  Having the hardare inhouse also means
> that the responsetime is less than 2 hours for complete replacement.
> 
> Along with the spares, I have standardized the configurations such that
> a complete machine failure would require just the removal of cables, the
> insertion of a replacement machine and a reboot.  Total time involved
> including travel to the datacenter:  2 hours max.
> 
> The replaced machine is "fixed" after replacement at the office and
> becomes the "hot spare".
> 
> The same procedure is used regardless of the actual failure.  Should a
> disk controller fail, the machine is replaced.  The controller is
> replaced later and tested again becoming the spare machine.

That is roughly the same approach we had in a third world country.  It 
was an EU funded project and that money was only available once.

Solution: buy double of everything and a stack of spares so that you 
always had one machine in working order; also consumables (yes, even line 
printer paper) to last you several years.  Give your own people adequate 
training to be self sufficient for both hardware and software maintenance 
over the anticipated lifespan of the project.

-- 
Paul Sture