I need a brief justification with estimated numbers for the time lost
in development for .NET/ SQL Server, etc, by having 256 MB of RAM vs
512K.

Basically, I need to justify the acquisition of more RAM to a manager
that appreciates the concrete (even if it's tough to quantify).

You could also approach this in a generic way for any environment, I
just listed the specifics as an example.

Any rough guestimation will be fine, as long as it's presentable.

---

Request for Question Clarification by maniac-ga on 17 Jun 2003 17:37 PDT

Hello Rohaab,

Getting to a "correct" answer to this may require a few steps on your
part.

[1] Know the cost to upgrade RAM (from 256M to 512M). This will vary a
lot - based on the type of system you have. I'll assume $100 for the
part and no cost to install; add a modest amount if you need to pay
for a technician. Use a site such as
  http://www.oempcworld.com/category.html?UCIDs=66526%7C1101982
for sample prices. Double the cost if there may be other factors to
consider.

[2] Know how much memory you currently use on a regular basis.
Assuming the use of Microsoft Windows, bring up the task manager and
see how much swap space you are using. Measure this when you have your
typical applications up and running. If the amount of swap used is
over a few hundred megabytes, I would ask for 1024 Mbyte instead of
512 Mbyte.

[3] Know how much your time costs the company. Alternatively, how much
more sales you can make if you are more productive. This is one area
your boss could say "but I won't pay you less if I give you this
memory" and be able to back it up (unless you are paid for overtime).

[4] Know how much money your boss has for computer upgrades. If there
is no budget for this - you are basically asking your boss to take the
money from some other purpose for buying this. There is a flip side to
this if you have money available at the end of the fiscal year - some
companies treat the budget as "use it or lose it".

[5] Know how much time you will save. Measure something simple such as
switching between windows when the task is swapped. In a similar
manner, if you can somehow measure compile time on a system that is
not swapping vs the swapping one would help as well. Saving a few
moments each hour will add up to the time savings to pay for the
upgrade. Since memory is so cheap, this should not take long, but I
always try to get a payback of one year or less w/ at least 2:1
benefit vs cost.

Then do the arithmetic. Put it in a table comparing costs vs.
benefits. If the analysis is right for you, it should be right for the
rest of the department / organization. If you do this for one person -
why not for 10 or 50? To justify a wider deployment, get one or two
machines upgraded and do a test - then go ahead with the rest of the
group. Depending on the company you may get a small award for
submitting a cost savings.

Is this the kind of answer you want or do you need something more
specific?

  --Maniac

I would definitely try to see how much ram your system actually uses
(or attempts to use) as described above. Really, only a very light
database server can usually get away with 256mb of ram. I've seen
database servers run by websites just to run their forums that require
dedicated dual processors and gigs of ram just to not screech to a
halt, and from my point of view, a forum really pretty simple in
nature.

With the cost of ram these days, I would recommend anyone using a
database for business functions take a close at how much ram their
system is using, and upgrade if appropriate, especially with only 256
ram in it. I know of a lot of people running databases where the
general practice is to put however much ram will fit in the machine,
because anything less will greatly decrease performance. Sometimes "as
much as possible" is 512 megs, but I've also seen servers upgraded to
up to 3 gigs of ram.

Really it all depends on the purpose, so if you could tell us more
about what this system will actually be doing, we could probably be of
more help to you.