Re: SQL server Vs Oracle

[Followups restored because I'll be responding to the non-inflammatory part of the post]

Stephen Harris (sweh_at_mpn.com) wrote:
: David (desertfox_at_thegrid.net) wrote:
:
: : anyway. Nice of you to 'trim' my original post but I was saying in that post
: : that the IT MANAGERS are the ones that are going to say 'who gives a rat's
: : ass to Word's paltry 100 meg install'.
:
: The IT managers with 500 client machines to manage, that's who. Not 386's
: but maybe a P133 with 1Gb disk - ie a machine under 3 years old and _still_
: not written off due to depreciation (some places run depreciation over 4
: years). I've got DX4-100's with 16Mb of RAM and 500Mb disk in use because
: they are still within "asset value" and can not be junked.
:
: 100Mb is now a relatively large install :-(

Any company that was buying DX4-100s with 500MBs of disk space in early 1996 was being extraordinarily foolish with their money. They should have been buying P133s with 2GBs of disk space (the machine I'm typing this on, a P120 with a 1GB HDD, was purchased in Jan. 1996 -- I was saving money by buying just behind the technology curve (120 instead of 133), and unwisely cut the default 2GB HDD back to 1GB. But I also upgraded the L2 cache to 512K, which is why the machine, now with 80MBs of RAM, is still a perfectly viable PC. And that's *with* Office97 Pro installed).

If you do have 1GB machines attached to a network and you don't have 100MBs of disk space free, those machines need to be cleaned up. There should be plenty of room on a machine like that *if* network storage is being properly utilized and *if* client software installation is being controlled.

But, again, you (and I) shouldn't have scrimped on the 1GB HDDs 3 years ago when those machines were purchased.

: : Microsoft applications on your 120 Meg drive and I will continue to develop
: : software applications.
:
: You _will_ lose a big market with that attitude.
:
: Businesses can't always afford the biggest and best machines. Inertia
: and sheer size prevents this. Realise that the home hobbyist probably has
: a machine twice as powerful as the typical office user.

Businesses cannot afford *not* to spend wisely. The machine with the lowest acquisition cost is often the most expensive in the long run. One should always buy workstations that will last 3 years without upgrades, or buy workstations that can be discarded after 1-2 years. The cost will be just about the same. If you're trying to cut cost below that, you're being penny- wise and pound-foolish. Over a 3-year lifespan, a $2,000 business PC costs $2.67 a day. A $1,200 PC, discarded after two years, costs about $2.40 a day. Add in a premium of the cost of installation/re-configuration/ re-installation (i.e., rebuilding the workstation every two years instead of every three), and the cost will be just about the same, while you've taken less of a performance improvement with the cheaper PC at the initial upgrade.

: : Upgrade if you are still running Win 3.1, and after you upgrade you will be
: : upgrading less and less than those people had to in the past. Is this such a
: : hard concept to understand?
:
: It's actually a falacy. People stuck with DOS for many many years. Even when
: DOS 6 was out, most apps still ran fine under DOS 3.3. The speed of
: releases of Windows versions combined with the speed of release of Office.
: A company that has just (eg within 2 years) rolled out Win95 are now looking
: at having to replace their Office95 systems with Office97 since their business
: partners are sending documents in this new format. Now they're going to have
: to look at Win98 or make the jump to NT, let alone looking at W2K etc.

Anyone who rolled out Win95 with Office95 two years ago, when there was already a new release of Office, was very foolish. It's important to time your upgrades to fall at reasonable points within the release cycle of your mainline software applications.

I've been advising clients for two years to evaluate NT as their next client workstation OS. This includes people who were still on Win3.x at that point. Some have chosen to go ahead with NT, others to migrate to Win9x, instead.

W2K is way out there in terms of client workstation OS. It's basically a version 1.0 OS, and anyone foolish enough to commit to it in a company-wide roll-out deserves whatever happens to them. I would say no one with any sense will be putting W2K on the desktop until the second half of 2001.

And anyone running Office95 at this point should be looking at Office2000, not Office97. Dell is already shipping with Office2K pre-installed.

: : I think I will. Memory is incredibly cheap. If you can't afford more memory
: : then you might consider another line of work as making money in the computer
:
: There are two issues here.
: 1) Server side. Yes, memory is easy and cheap. Installing more memory in
: your SQL Server is always viable - to the limits of the machine. However
: NT doesn't always properly manage memory resources, so this will aid
: stability, but not solve leak problems

If you spec'd out the server to the proper level of power when you purchased it, you shouldn't have any problems.

: 2) Client side. See above discussion on how to roll out memory expansions
: to 500 desks.

An upgrade of a 32MB client workstations to 96MBs could get you another extra year of use (for a P166 or faster). Cost: $75/each for DIMMs, $130/each for SIMMs, 20 minutes of work for a tech (if internal, that's less than $20, external techs ought to be on the order of $25-30), for a maximum total investment of $160. That's $.64 a day for the additional year of productivity.

Of course, I'd say this should have been done sooner, rather than later. Three years ago, 32MBs was the standard "power" memory configuration. One year later, it was 64MBs (with the release of Office97). If the upgrade had been done then, the benefit would have lasted even longer, and the cost per year of the upgrade would have been less. Further, at that point SIMMs were much cheaper than they are now, so it would have cost less in absolute terms.

The key point here is that money well-spent will return in value, productivity and performance many times over the original investment. Cut corners, nickle and dime it, and you'll end up spending many times over the original investment just to keep from falling behind, while not getting the benefit of the better hardware until it's already barely able to cope.

Penny-wise, pound-foolish.

--

David W. Fenton                        http://www.bway.net/~dfenton

dfenton at bway dot net                http://www.bway.net/~dfassoc