> And, a 100 megabit LAN is, indeed, pretty good at delivering data. But, if
> you are using a front-end that is stored on another machine, it will also
> have to deliver every object in that front end that you load, to each user
> who uses it.
> > Access will be writing/saving internally
> > in the front end. Having multiple users
[quoted text clipped - 6 lines]
> to the linked table database ( BE ). What
> would be the nature of these writes and saves?
Access' internal working storage... that is not written to tables in the
linked database. Like every other Windows program, it obtains memory and
writes its own files (not tables) that it uses in doing its work.
> Wouldn't the read-only status of the FE
> prevent this ? Or doesn't windows read-only
> really work.
No, I don't think you can successfully specify an Access front-end to be
read-only and have it work.
> > And, a 100 megabit LAN is, indeed,
> > pretty good at delivering data. But, if
> > you are using a front-end that is stored
> > on another machine, it will also have to
> > deliver every object in that front end
> > that you load, to each user who uses it.
> True. But don't fequently used objects
> end up in paged memory on the FE
> instance machine?
Nothing can be cached in that manner until it is retrieved. And, I am not
knowledgeable about the performance improvements due to memory caching. I do
know that a 100 MBPS network is a poor substitute for a local disk drive.
> We've been operating in this environment
> for 7 years, and forms with query data
> seem to have response times of 100-200 ms.
Response times of 100-200 NANOseconds? I wouldn't even know how to begin
measuring that.
I haven't had any experience, nor heard any reports, on performance over the
new gigabit networks, but one of those would be a much better substitute for
a local hard drive.
In any case, the vast preponderance of opinion is that each user should have
their own copy of the front end.
> BTW, Thanks Larry, for your serious
> and thoughtful responses.
You are welcome.
> I may be trying to cary over too
> much of my background in
> mainframe DB2 and SQL/DS,
> VM, MVS, TSO, and VS/APL.
There are certainly very major differences in mainframes and in networked
PCs. In a previous incarnation, I, too, was a mainframer and minicomputer
programmer.
Larry Linson
Microsoft Access MVP
Ed Mulock - 06 Dec 2004 15:30 GMT
> > > Access will be writing/saving internally
> > > in the front end. Having multiple users
[quoted text clipped - 10 lines]
> linked database. Like every other Windows program, it obtains memory and
> writes its own files (not tables) that it uses in doing its work.
Well, Access isn't suposed to be every other Windows program !
It's suposed to be a multi-user application. If it writes its own files, it
should keep track of the user ( or better yet the transaction) so it doesn't
corrupt them.
> > Wouldn't the read-only status of the FE
> > prevent this ? Or doesn't windows read-only
> > really work.
>
> No, I don't think you can successfully specify an Access front-end to be
> read-only and have it work.
You can. And, it works( in VERY limited testing). You can open, update and
insert into linked tables.
> > > And, a 100 megabit LAN is, indeed,
> > > pretty good at delivering data. But, if
[quoted text clipped - 11 lines]
> do
> know that a 100 MBPS network is a poor substitute for a local disk drive.
It's definately slower. But poor or good substitute depends on your
requirements.
I happen to think distributing code all over the user base is a poor
substitute for having a shared code store.
> > We've been operating in this environment
> > for 7 years, and forms with query data
> > seem to have response times of 100-200 ms.
>
> Response times of 100-200 NANOseconds? I wouldn't even know how to begin
> measuring that.
"ms" is a milisecond, not a nonosecond. 200 ms is .2 seconds.
> I haven't had any experience, nor heard any reports, on performance over
> the
[quoted text clipped - 5 lines]
> have
> their own copy of the front end.
I'm aware of the prevalence of the "opinion". I'm questioning it.
> > BTW, Thanks Larry, for your serious
> > and thoughtful responses.
[quoted text clipped - 9 lines]
> PCs. In a previous incarnation, I, too, was a mainframer and minicomputer
> programmer.
And I'll bet you'd go ballistic if someone corrupted a multi-user
application by creating a scratchpad work area that mixed up the data from
multiple users.
> Larry Linson
> Microsoft Access MVP
Larry Linson - 09 Dec 2004 00:53 GMT
Ed --
I have had no reason to question that having multiple users logged in to the
same front-end or monolithic database does increase the chances of
corruption. I have seen and read of too many instances to doubt it.
There's not being overly happy with the way some software is desinged and
implemented; there's also being so unhappy with it that you feel compelled
to do the equivalent of spitting into the wind. That, at least, I have
learned in my years (since 1958) in the computer business. And, I try to
limit myself to the first and go ahead and work with the software rather
than against it.
I have been splitting front-end and back-end since the days of Access 2.0,
but I acknowlege that Access 2.0 seemed to be a lot less sensitive to having
multiple users logged in to the same front-end.
I've had no troubles (also since Access 2.0 was a stripling) with
determining when there was a need and informing the user that they could or
must download a new copy of the front-end. I believe you'll find a writeup
at http://accdevel.tripod.com.
MVP Tony Toews took a different approach and created an Auto FE Updater,
which you can find at his site, http://www.granite.ab.ca/accsmstr.htm.
Those approaches ought to take some of the pain out of distributing front
ends to multiple users. Good luck with what you are trying to do.
Larry Linson
Microsoft Access MVP
>I was not aware that an Access FE would be writing/saving data to itself,
>rather than to the linked table database ( BE ) . What would be the nature
>of these writes and saves. ?
One source is query plans. Another would be form filters and such.
I'm sure there are other things of which we're unaware.
>Wouldn't
>the read-only status of the FE prevent this ? Or doesn't windows read-only
>really work.
I recall reading one or two postings a while back stating that keeping
the FE as read only did work for him. Presumably though he wasn't
relinking tables or adding temporary linked tables as required and
such.
>> And, a 100 megabit LAN is, indeed, pretty good at delivering data. But, if
>> you are using a front-end that is stored on another machine, it will also
[quoted text clipped - 3 lines]
>True. But don't fequently used objects end up in paged memory on the FE
>instance machine ?
One of my clients is running an terminal server system and placing the
users FE on the file server. It's working for them but we've never
tested the difference in response times doing this. I suspect it's
slightly sluggish but the IT admins didn't want me to even test this
scenario.
Tony
--
Tony Toews, Microsoft Access MVP
Please respond only in the newsgroups so that others can
read the entire thread of messages.
Microsoft Access Links, Hints, Tips & Accounting Systems at
http://www.granite.ab.ca/accsmstr.htm