How much is too much for Access?

[Jacqui]

I'm looking for documentation, articles or other
published/authoritative sources discussing recommended
size limits for an Access database back-end - I know that
for v2 the size limit for an .mdb file was 2GB and assume
2K and XP are the same(?) But then there is always this
question of whether the amount of data Access can
adequately "handle" - So how large is too large for
Access? How to address those vague "scalability" concerns
and settle a debate over whether it's ok to develop this
with an Access back-end or go to (and this is often
overkill) SQL Server (or mysql). This comes up on almost
every application I design. I can't believe I can't find
something tangible on this in the knowledgebase

 

[Jeff Boyce]

Jacqui

I suspect the reason you have not found a hard and fast "rule" is because of
the number of variables to be evaluated. For instance, how many folks will
be using the application? In what manner (data entry vs. simple lookup vs.
....)? Over what speed network? With what kind of PCs/horsepower/RAM/...?

And the application itself -- are the forms bound to the entire recordset,
or are they unbound, using "marshalling" routines to get and send data?
And... And... And...?!

Take a look at a Google search on this topic for further descriptions of
factors, but please post back here if you find someone who presents a
rationale/rule for deciding! I'd be interested in that.

 

[TC]

It reminds me of an old(?) project estimating technique called Function
Point Analysis.

With FPA, you estimate the #s of tables, screens, reports, inputs, outputs,
processing rules & so on, and plug these numbers into a big formula. This
gives you a single numeric result - the # of "function points" required for
the project. Then you just multiply that # by 'x' - the magic value
converting function points to hours of work!

Of course, no organization has any idea, what is their value of 'x'. In my
experience, a reasonable estimate can usually be obtained by taking the
project leader's telephone number, multiplying by the # of pregnant women in
the organization, and subtracting the outside temperature in degrees
centigrade.

Yet another doomed attempt to make a complicated issue, simple!

Cheers,
TC

 

[John Vinson]

I suspect the reason you have not found a hard and fast "rule" is because of
the number of variables to be evaluated.

Agreed. Another key variable is user expectation. I've encountered
users who were amazed and delighted at fifteen-second response time on
queries ("it's so FAST!!!"), and others who were ready to junk the
project because of a perceptible half-second pause before seeing
results.