[ale] little rant

Lightner, Jeff JLightner at dsservices.com
Wed Apr 1 16:39:37 EDT 2015


Yeah but if you already did the development to share disks in a cluster is it really easier to rewrite so you no longer share than to bring that part of the code forward?

My own view isn't that it is cheaper but that many developers are lazy and sloppy.   This is one reason so many of them squirrel away code in their home directories rather than bothering to use source code control properly and therefore end up reintroducing bugs fixed in earlier releases into later ones.   They take their own copy from before the bug fix, then when developing something new push it back out.   I'd like to think this was restricted to internal coding but the number of 3rd party and/or commercial products (including OSes) I've used that reintroduced previously fixed bugs in later versions shows it isn't.

At one job I got so fed up with having to "fix" a "broken" confg file so often that I finally created a cron job that periodically deleted everything older than 1 month in developers' home directories.    Since none of them could ever admit they'd been doing things without checking them in I never got a single complaint.

-----Original Message-----
From: ale-bounces at ale.org [mailto:ale-bounces at ale.org] On Behalf Of DJ-Pfulio
Sent: Wednesday, April 01, 2015 3:56 PM
To: ale at ale.org
Subject: Re: [ale] little rant

It isn't just Exchange.
Zimbra used to require 1G of RAM.
Then 1.5G
Then 2G
now the current release has 8G as the minimum required (not true, but ...)

Back in the olden days, supported hardware was very restricted for Linux. When you only support 5 pieces of hardware, being smaller is easier.

I'm not trying to say that current OSes aren't bloated, just that it matters less and less as cheap HW is available.  SW dev time is much more expensive than HW at this level - unless the device is consumer and has 20M installations. I've done some similar math and found it frustrating how expensive bad software can be.

On 04/01/2015 02:59 PM, Lightner, Jeff wrote:
> One of the things I learned early in the PC days was that every 
> hardware advance to increase resources to help software performance 
> was always followed by "enhancements" to the software that took every 
> bit of new resource and begged for more.
> 
>  
> 
> One of my favorites was MS Exchange 2010 that decided you should 
> assign as much space to each node of a cluster (i.e. duplicate space) rather than share the
> same SAN space on both nodes as earlier Exchange clusters did.   Why "Because
> disk storage has gotten cheap". 
> 
>  

_______________________________________________
Ale mailing list
Ale at ale.org
http://mail.ale.org/mailman/listinfo/ale
See JOBS, ANNOUNCE and SCHOOLS lists at
http://mail.ale.org/mailman/listinfo



More information about the Ale mailing list