3-tier winforms architecture

[David Noble]

We've been developing a web site using 3-tier architecture for 18
months now.
There is a common layer that defines the classes - using XML schemas.
The data layer acts as a wrapper to 3 databases - SQL Server, Oracle
and AS400.
The business layer exposes web services which communicate with the
front end, ASP.Net. All 3 tiers are on different boxes. This works
well.

Now I am leading a team to build a winforms app. I need some advice as
to the best way to proceed. The application will have 9 or 10 users
only.

Are there any 3-tier examples demonstrating remoting, or are web
methods the way to go. I am anticipating a lot of deployments after we
go live, so I want to keep as much as possible on the server, and not
have to go round re-installing every couple of days. Or are Web
Services still the way to go ?

How does deployment work ? Could I put a client exe on the user's PC
and remote in to the business tier which talks to the data tier on the
same box ?

Can anyone point me to some good tutorials to help ?

 

[Jan Elbæk]

About deployment of Windows Forms Apps. Please take a look at Smart Client
Application Model:
http://msdn.microsoft.com/netframework/using/building/windows/analystreports/smartclient.aspx

And on remoting. Pick up Ingo Rammers book: Advanced .NET Remoting. At Ingo
Rammers web site theres an excellent .NET Remoting Use-Cases and Best
Practices http://www.ingorammer.com/RemotingFAQ/RemotingUseCases.html

/Jan

 

[Guest]

***NOTE: Don't take this message personal, it mainly directed to the
stupidity of n-Tier idiots and the ignorance of the Microsoft architecture
team and those gurus and authors who regurgitate this nonsense without ever
questions the real business need and value and PLUS actually do an ROI over
the different architectures...they are still stuck in this Windows DNA
model....plus....it's only for Microsoft's Windows applications
anyway....Microsoft itself.....not you the customer as Microsoft = Windows
which means please develop the architecture so Windows will still have some
usefullness to the common surfer on the Internet.
***

18 Months for 3-tier!!!! Shows how incredibly weak n-Tier is.
That's on par as SAP....any project, big or small going over a 9-12 months
is out of touch with the real world.

Now, you are going to make a Windows Apps for only 9-10 users?

For what reason? Just what business advantage will you achieve making a
Windows App?

Rich client??? Like what is it in that windows app that can't be done (and
is worth doing) on a Web app.

What happens if a few of those 9 to 10 people say, "Hey I really need this
on the Web, Or I need to access this somewhere else, like in the conference
room, which doesn't have the software installed, or we don't want it
installed there".

People think that that Windows App give a RICH CLIENT, but just exactly what
is the BUSINESS Advantage that a Windows Client will give you? Unless you
need real time access, there is absolutely no need for it other than to
impress your boss with some pretty pictures...after a while, your boss will
NOT use it as it will be too inconvenient and of very little business value
to him/her

Now, as I read, you have 3-tiers on THREE different boxes!!! Shows the
confusing nature of 3-tier to think that you need 3 boxes to run it.

If there is a long running application, stick all that logic on it's own web
server and combine that with the web logic. So you have 2 web servers and
one db server.

By using 3 different boxes, you only made twice as much work for yourself as
it took you 18 MONTHS!!!
Plus, you have this MIDDLE tier that is a single point of failure and of
which gives you lower performance in the first place and now there is 2 hops
to the presentation tier. Furthermore, now you have code in 3 different
places which will be great for debugging and performance tuning!!!

You got this 3-tier thing built in order to swap out data tiers...well guess
what, you already target all three databases anyway, so you essentially
defeated any need for a database tier anyway. You are not swapping in and
out the SAME stuff from Oracle to Sql Server every other day...so it's
pointless to design for this....Data from the Oracle DB is going to stay or
come from the Oracle DB and the same for the Sql Server and AS400.

Now, if you say you can easily switch, you should have used some of those 18
months to do the switch instead of building for the *possibility* of
switching.

But how many people are going to re-write all those stored procs and redo
all those tables when it's already running on the db's? zippo!!!

 

[Eddie de Bear]

Obviously you have never worked on a large project...
Quite often, clients will require large applications to
be written, either web or rich client.. Implement the
application in phases, each phase may take 2-3 months for
dev/test/release cycle, but the whole application (or
system) could easily take 18 months..

Why a windows app.. Simple.. Maybe the client has an
account package or something similar that they need to
integrate into an application.. This sort of integration
is far better done with a windows app (or linux desktop
app..)

As for 3 tiers on three different boxes.. Once again, you
have obviously never worked on a large project. Have you
heard of scalability??? If one of the tiers is performing
badly, then you can very easily throw in more boxes...
Which also has the advantage of removing the single point
of failure.. That is the advantage of the three tiered
architecture...

Debugging, not an issue, use the .NET remote debugging,
it's awsome... Works a treat..
Performance monitoring... Once again, not an issue, Throw
in performance counters, traces, logging and your all
set... It's really not that much of an issue, you just
have to understand debugging outside of the IDE...

I don't think he was talking about swapping the
databases... I think you'll find the application is
actually using data from 3 different datasources at the
same time.. Once again, sounds like you have never needed
to do that either... The best thing about this approach,
is the business layer does not need to worry about which
of the three datasources the data is comming from... Is
just uses it..

I hope this has helped in you understanding of N-Tiered
development...

Best Regards

Eddie de Bear
MCSD

 

[Steve S]

WELL...... that was a real pile of $H!T....... guess they never really
heard of extending a component to expose different interfaces..... some of
them as EnterpriseService components..... and others via WebServices......
and yet otheres via RemoteServices..... and it all operates via the same
common code..... running below the covers.......

User the appropriate methods and parameters..... to gain performance.....
keep things thin.... when crossing many hops...... push stuff via XML
when you need to jump through firewalls.... but binary for speed
inside...... where you probably will be moving large amounts of data......

DUH !!!!!!! Talk about being out of touch ? components...... SMALL....
easy to implement..... not an issue....... and if I build the component
for 9 folks today...... tomorrow.... 90 could want that same
capability... IN SPADES !!!!

Keep your options open...... do the n-tier design..... don't fall into
the trap...... of embedding all the stuff into ASP.... as that limits your
deployment options...... and ability to extend......

Data-tier ? real kewl stuff..... ours works just fine..... and our
programmers only write SELECT statements..... the data-tier does all the
rest...... OH YEA...... no one writes our insert-update-deletes......
not even the WIZ.... as our experience is that the stored procedures might
be a little quicker..... than dynamic SQL..... but the expertise required
to create..... and MAINTAIN those..... is not worth the effort/price
difference...... then again... we didn't spend months building out
stored procs.... just several days..... writing the Selects......

Oh yea.... that same data tier... does allow us to automatically cut
over..... to a back-up database...... if we can't contact our
primary...... but then again..... of what value is the datatier..... ?

I totally agree that three boxs is a little much..... when all of our runs
on a cluster..... for the business and data layers...... the GUIs.....
is run through a terminal server farm..... so we have TWO sets of
clusters..... with about 500 users nationwide....... running several
applications......


It isn't all of us that need to live with YOUR decisions........ you will
find a complete spectrum of answers...... N-Tier works...... and works
well...... how long it takes..... depends on your ramp time.....

 

[Guest]

If what you say is sooo true, why do 60% of all CIO think that IT sucks.
If what you say is sooo true, why did McDonald's fail.
http://www.baselinemag.com/article2/0,3959,1184848,00.asp

Try not to argue as their was an argument before, and all I had to do is
simply point out all the failures that are OH SO published on the
WEB.......For every published failure, there is about 1000 more unpublished.


Keep your options open? You are the one who is so closed to simplicity.

Data-tier ? real kewl stuff..... ours works just fine..... and our
programmers only write SELECT statements..... the data-tier does all the
rest...... OH YEA...... no one writes our insert-update-deletes......
not even the WIZ.... as our experience is that the stored procedures might
be a little quicker..... than dynamic SQL..... but the expertise required
to create..... and MAINTAIN those..... is not worth the effort/price
difference...... then again... we didn't spend months building out
stored procs.... just several days..... writing the Selects......

PLUS, YOU think you are SOOO smart, that you don't even know SQL. I know
SQL.
BUT you don't! Very ODD....seems more like stubborness.

I can easily write SQL and it's better for a number of reasons, but then you
come right back around and say n-Tier is sooooo great and stored procs suck.
How ironic, Mr. N-Tier.

You don't know crap about enterprise systems.

See INLINE BELOW for more of my comments.

WELL...... that was a real pile of $H!T....... guess they never really
heard of extending a component to expose different interfaces..... some of
them as EnterpriseService components..... and others via WebServices......
and yet otheres via RemoteServices..... and it all operates via the same
common code..... running below the covers.......

User the appropriate methods and parameters..... to gain performance.....
keep things thin.... when crossing many hops...... push stuff via XML
when you need to jump through firewalls.... but binary for speed
inside...... where you probably will be moving large amounts of

data......

WHOA!! wait a second....here we are talking about performance, but below you
say don't use Stored Procs!
Yet with Stored Procs don't need to pass that much data around as a SQL
Query? Network bandwidth, right?

You are writing so much code but then you can't even write INSERT and UPDATE
stored procs.....that's like saying your can add and multiply, BUT you can't
substract and divide.

 

[Guest]

COMMENTS INLINE FOR THE THESE SO-CALLED ENTERPRISE ARCHITECTS

Obviously you have never worked on a large project...
Quite often, clients will require large applications to
be written, either web or rich client.. Implement the
application in phases, each phase may take 2-3 months for
dev/test/release cycle, but the whole application (or
system) could easily take 18 months..

Ahhh yes, the SLOW antiquated dev/test/release method....

****************************************************************************
**
OH by the way, in the REAL world, all of the TOP and BOTTOM IT firms skip
the "TEST" cycle cause they are SOOOOO FAR behind in the schedule, they will
eventually lose the contract if they don't DELIVER something.....
****************************************************************************
**

That's the REALITY......

any IT project taking over 12 months is stupid as the design specs have
already changed in the business.

Why a windows app.. Simple.. Maybe the client has an
account package or something similar that they need to
integrate into an application.. This sort of integration
is far better done with a windows app (or linux desktop
app..)

Of what REAL value is this Windows client to the END Customer when it can be
done on the Web?

As for 3 tiers on three different boxes.. Once again, you
have obviously never worked on a large project. Have you
heard of scalability??? If one of the tiers is performing
badly, then you can very easily throw in more boxes...
Which also has the advantage of removing the single point
of failure.. That is the advantage of the three tiered
architecture...

Wrong!!...if one of the tiers is BROKEN, you don't replace it with another
copy of what's BROKEN
Performing BADLY and being BROKEN have their DIFFERENCES.

One of the biggest reasons it's performance badly is because the thing is so
OOP'd to death and so-called COM'ed out, it's creating object for even the
simpliest of tasks.

Then, cause you got this object calling this (1)object and this (2)object
and this (3)object and this (4)object and this (5)object, you eventually
FORGOT what you wanted to do in the first place....HA HA HA....Like you
really know how to do Enterprise systems.......

Debugging, not an issue, use the .NET remote debugging,
it's awsome... Works a treat..
Performance monitoring... Once again, not an issue, Throw
in performance counters, traces, logging and your all
set... It's really not that much of an issue, you just
have to understand debugging outside of the IDE...

Performance is not that easy after you done the coding. You still have to
RE-CODE again to fix all that crappy n-Tier architecture in the first place
SO in the end, you DON'T have n-Tier architecture....oh well, you just have
to learn the hard way.....

I don't think he was talking about swapping the
databases... I think you'll find the application is
actually using data from 3 different datasources at the
same time.. Once again, sounds like you have never needed
to do that either... The best thing about this approach,
is the business layer does not need to worry about which
of the three datasources the data is comming from... Is
just uses it..

The business layer is a load of crap cause no one changes databases in the
first place for the same business functionality.......If you are going to
change the database, you are going to re-write the sprocs anyway on this
database port....you are basically designing in this modularity at the cost
of performance as well as reliability of the code in the first place.

The database modularity feature is a feature that has SO LITTLE BUSINESS
VALUE as it doesn't happen that often in the business world in the first
place.......I don't know of anyone who goes out and spends money on Oracle
and then switches to DB2 and throws out the Oracle License.

.....if anything, it's more cost efficient to buy faster hardware, keep the
license that you can't sell back to Oracle in the first place and install
the new server. That, in the end, is more cost efficient. It's called use
what you just purchased and can't return.

But let's say they switch to SQL Server from Oracle....now what are they
going to do with that Oracle DB then? They just paid tens of thousands for
it. Now you are saying, "Let's build a data-tier to be able to SWITCH BACK
from SQL Server to Oracle." That's not one, but TWO switches!!!

That's LUNACY...make up your mind!!!!


If you are going to make a switch, one of the things in deciding whether to
switch is going to be performance in the first place. And if it's
performance, why do you turn right back around and advocate a much less
performing n-Tier architecture????

At first, you say the reason to switch is due to performance(or size), so
you go from SQL to Oracle. Now you say it's COST/PEFORMANCE, as you can get
two SQL Servers for the price of one Oracle. So you switch back!

The cost of the PORT is going to gobble up all the so-called cost savings of
this "so-called" modular data tier in the first place cause IT's NEVER that
easy..

And if your are big enterprise, they already have their code in place so
they are not going to need Mr. IT shops
modular-code-that-can-work-with-any-database, as they already picked a
database and have it running in production anyway.


Please stop with these silly arguments as I can easily use Microsoft OWN
marketing slogan against itself and I havent' even mentioned that one.......

See this threads....this guy lost big time.....

You as well as all MVP's, MCSD, Gurus have soooo much to learn.

http://groups.google.com/groups?q=R...=#[email protected]&rnum=1


JUDGEMENT DAY FOR OOP COMPLEXITY Re: Reasons for a 3-tier ...

http://groups.google.com/groups?hl=...3F+Why&meta=group%3Dmicrosoft.public.dotnet.*


Hey eBay..Let's talk about OOP and Reliability.........Re: Reasons for a
3-tier achitecture for Web? Why

http://groups.google.com/groups?hl=...3F+Why&meta=group%3Dmicrosoft.public.dotnet.*


Now, ask yourself why do MOST IT projects FAIL?

What's so similar in all these big projects?

Well what's similar is you are all using this same silly n-Tier OOP design
that you learned in Windows and are applying it to the Web.....PLUS, you
think that people want BOTH, Windows and WEB....wrong, they don't and can't
afford it in the first place.


Here, another FAILURE

http://www.baselinemag.com/article2/0,3959,1184848,00.asp

 

[David Noble]

18 Months for 3-tier!!!! Shows how incredibly weak n-Tier is.

That's on par as SAP....any project, big or small going over a 9-12 months
is out of touch with the real world.

I feel I have to respond to this rant. 18 months ? I should also tell
you that we've gone from 1 up to 15 developers working on this at any
one time, with a core of 10! This is because the business is
incredibly complicated. To give you an idea, the Oracle system alone
holds 3500 tables, and the business logic behind the AS400 system is
40 years old!

On the hardware side, by using 3-tier we've been able to site our web
servers externally, at an ISP, and have our app servers in-house, both
as part of web farms. If we need to scale out, this in now easy. The
business is time-critical. We can't just ask the customers to wait. We
just buy another server and add it to the cluster.

Now, you are going to make a Windows Apps for only 9-10 users?

For what reason? Just what business advantage will you achieve making a
Windows App?

OK. Part of the system includes temps swiping in as they turn up for
work and leave. I don't believe you can do that through a browser. And
it has to work standalone, when the network fails.


Rich client??? Like what is it in that windows app that can't be
done (and

is worth doing) on a Web app.

What happens if a few of those 9 to 10 people say, "Hey I really need this
on the Web, Or I need to access this somewhere else, like in the conference
room, which doesn't have the software installed, or we don't want it
installed there".

People think that that Windows App give a RICH CLIENT, but just exactly what
is the BUSINESS Advantage that a Windows Client will give you? Unless you
need real time access, there is absolutely no need for it other than to
impress your boss with some pretty pictures...after a while, your boss will
NOT use it as it will be too inconvenient and of very little business value
to him/her

So you'd always build a web app no matter what ? Winforms does give
you a richer UI. And if we used dot net remoting, you could use it
away from the office. And if we use a smart client, deployment is easy
too. You should read the articles that were recommended in the
previous post.

Now, as I read, you have 3-tiers on THREE different boxes!!! Shows the
confusing nature of 3-tier to think that you need 3 boxes to run it.

Actually 10 boxes. But we could scale out the app farm or either of
the web farms.

2 external web servers,
external state server,
2 internal web servers,
2 app servers,
1 SQL Server,
1 AS400,
1 Oracle

By using 3 different boxes, you only made twice as much work for yourself as
it took you 18 MONTHS!!!

And it's not finished yet! We've got another 2 years of work in the
pipeline.
The business itself has just started to take off because we've put it
on the web!

Plus, you have this MIDDLE tier that is a single point of failure and of
which gives you lower performance in the first place and now there is 2 hops
to the presentation tier. Furthermore, now you have code in 3 different
places which will be great for debugging and performance tuning!!!

That's why we've got 2 app servers. So if one fails, we can swap it
out without too many wories..

You got this 3-tier thing built in order to swap out data tiers...well guess
what, you already target all three databases anyway, so you essentially
defeated any need for a database tier anyway. You are not swapping in and
out the SAME stuff from Oracle to Sql Server every other day...so it's
pointless to design for this....Data from the Oracle DB is going to stay or
come from the Oracle DB and the same for the Sql Server and AS400.

Now, if you say you can easily switch, you should have used some of those 18
months to do the switch instead of building for the *possibility* of
switching.

But how many people are going to re-write all those stored procs and redo
all those tables when it's already running on the db's? zippo!!!

We could have redesigned the database around a new SQL Server, but
that really would be a large project.

I guess you presumed it was just me taking my sweet old time.

 

[Steve S]

The mouse from the corner... slides the thumbnail over to the leg of a
chair... gets on top... and clears it's throat and takes a sip of
water... and begins to speak...

Let me make this VERY CLEAR........ Your arguements..... are as hollow as
your brain....... most of the CIOs believe what their directors tell
them...... and most of them came from big blue...... they also need to
stop and smell the roses.....

If what you say is sooo true, why do 60% of all CIO think that IT sucks.
If what you say is sooo true, why did McDonald's fail.
http://www.baselinemag.com/article2/0,3959,1184848,00.asp

As for your link........ LMAO.... you should read the reference before it
makes your point...... and I quote.....

In 1991, after looking at commercially available point-of-sale registers
from vendors such as Panasonic and PAR Microsystems, Dill says, McDonald's
decided to go it alone and have systems integrators build the register
terminals for its U.S. stores from off-the-shelf PC and peripheral
technology.

[ 1991... LOL and your using this to make an arguement for today's systems
?]

As a result, McDonald's was able to hold the cost of POS systems down to
roughly $15,000 to $20,000 per restaurant, according to Dill, cutting the
cost in half. [ sounds like a success to me ]

The software McDonald's developed, called PC POS, is a two-part system. The
actual point-of-sale terminals at the counter and drive-through window run
on software written for Microsoft MS-DOS. It provided "software
functionality for cash registers" says Dill, "like taking orders,
communicating to cooking operations and giving change." [nice work for such
a limited OS of the day]

PC POS and the in-store processor did manage to give McDonald's its first
real grip on its supply-chain data-and with the cancellation of McDonald's
Innovate project, it remains as the backbone of McDonald's system for
collecting operations data from its restaurants. [WOW.........]

OK... enough of the quotes from your FAILURE STORY.......... I have seen
enough of this failure to fail story.....

Keep your options open? You are the one who is so closed to simplicity.
PLUS, YOU think you are SOOO smart, that you don't even know SQL. I know SQL.
BUT you don't! Very ODD....seems more like stubborness.

I can easily write SQL and it's better for a number of reasons, but then you
come right back around and say n-Tier is sooooo great and stored procs suck.
How ironic, Mr. N-Tier.

If I have a very smart DAL.... I can even get my junior developers into the
act of writing SQL code and business logic... I can use my real wiz kids
for the important things... like mentoring the young ones... on how to
write good applications... [and like a few other folks on this NG, you
failed to read the message, just blew out smoke]... the TCO is much better
on SQL based systems... vs stored procs..... OH YEA... does that title
come with a NameTag and a raise ?

WHOA!! wait a second....here we are talking about performance, but below you
say don't use Stored Procs!
Yet with Stored Procs don't need to pass that much data around as a SQL
Query? Network bandwidth, right?

LMAO... you are a real class act....... increased bandwidth ? What a
total joke...... after the call into the database with the SQL call.....
how much traffic is there until you get the results ? how many hops is it
going to take... and... how are you going to get the databack.. if it
doesn't move data around ?

OH HELL.. let's get real to the point......... how many linear feet of
cable is it going to travel through at what bandwidth ? and compair that
to the traffic required for a 5 parameter stored proc....... and when you
finished...... come back and post the results...... PLEASE ENLIGHTEN
US........

The real benefit of stored procs is it is already bound to the database...
the DB as already determined the access path it is going to take and
optimized the request for the indexes... so for each call you make into
the database, with dynamic SQL you are paying a few extra milliseconds
simply because this extra work needs to be performed..... once it gets past
there... it is all the same...... but then again..... you already know
this..... and figured it was way beyond our technical ability to handle
this extreme knowledge.....

You are writing so much code but then you can't even write INSERT and UPDATE
stored procs.....that's like saying your can add and multiply, BUT you can't
substract and divide.

I never said I couldn't write them....... Why would you bang on a boulder
all day long with a small hammer... just to break it.... when you can do
the job in 30 seconds with C4 ? EFFECTIVE.... but messy.....


You have your opinion.... and others have different ones..... they all
can live in this world together...... without issue......

Yoiu obviously have a very high opinion of yourself.... but you also make
some very bold statements about me... and other folks.... and don't even
have the balls to post without hiding behind an alias.... almost like your
afraid to be known.... Do you have an identity crisis going on ? I know
a good doctor for that...... [oh shut up.... you don't know a thing] {and
who the F*&K asked you ?}

Take some of that BS you posted..... and throw it on your roses......
maybe they will bloom..... and give you something better to smell than that
BS......

The mouse steps down from the thumbnail.... and slides it back to his
mouse hole...... Excuse me..... but all this discussion has made me a
little hungry..... I think I will go cut me some cheese....... (as it
snickers........)

 

[Eddie de Bear]

Dude,

Really... I've been working on N-Tier architecture for
several years now, as well as deploying that architecture
across various numbers of machines. (Sometimes all
deployed on one machine, other times deployed across
10+)...

What I have learned over the years is very simple.. Some
N-Tier projects fail simply because the developers that
are selected for the projects cannot understand simple
programming.

Yes, I'll admin there is a tendency to "Over OOP" an
application, that can be an issue, but that normally
comes from inexperience... N-Tier and OOP are still
fairly new to many developers/architects, and as such is
miss used way to often...

Don't judge a programming methodolody on it's failure..
There are always failures.. Learn waht you can, try them
for yourself, and have a good look at how many Success
Stories there are.. You'll probably be very supprised...

You say "What happens if a tier breaks.. " simpe.. Fix
it... What happens if you single layer application
breaks.. easy.. Fix...

And yes, project do quite often take > 1 year for
development... Imagine a government department, that
always has changing requirements... The project will
never finish.. that's why government departments
generally have very large IT departments.. Laws Change,
regulations change, that's something we have to deal
with... The N-Tier achitecture helps handle these
changes, without having to root through a single project
consisting of thousands and possibly millions of lines of
code...

As for windows forms, what can you do for a client that
you can't with web.. simple... Try implementing a project
management system that allows dragging around resources,
try getting a user to select a single record form a large
set of data (It's not always possible to do with a drop
down combo ).. What about Imaging...

Yes, I am well aware of all the technologies that you can
use to do this sort of thing using a web based client,
but true usability is in windows apps...

Lets also talk development methodologies...

DEV/TEST/Release.. yep, it's old, so what.. it works...
To many companies ignore the testing, they release their
apps and guess what.. The app falls on it's arse,
management yells, and the project gets cancelled... then
someone mentions it was an N-Tier project... Next thing
you know, there is a case study on the internet about how
bad N-Tier is, all because management decided it was to
costly to do testing...
Testing is required, it's not an option.... TEST TEST
TEST...

You don't need to follow the strict DEV/TEST, Partial
testing can be done while still in development. Unit
testing can be done, and some usability testing can be
done.. It really doesn't have to cost alot...

I think maybe you need to move around a little more, do
some contracting, get some more experience, and get out
of the little two bit company that you work for.. Because
obviously it hiding you from the real world..

Last but not least, don't believe everything you read on
the internet, especially if it's someone elses opinion on
slashdot

-----Original Message-----
COMMENTS INLINE FOR THE THESE SO-CALLED ENTERPRISE ARCHITECTS





Ahhh yes, the SLOW antiquated dev/test/release method....

********************************************************* *******************
**
OH by the way, in the REAL world, all of the TOP and BOTTOM IT firms skip
the "TEST" cycle cause they are SOOOOO FAR behind in the schedule, they will
eventually lose the contract if they don't DELIVER something.....
*******************
**

That's the REALITY......

any IT project taking over 12 months is stupid as the design specs have
already changed in the business.




Of what REAL value is this Windows client to the END Customer when it can be
done on the Web?



Wrong!!...if one of the tiers is BROKEN, you don't replace it with another
copy of what's BROKEN
Performing BADLY and being BROKEN have their DIFFERENCES.

One of the biggest reasons it's performance badly is because the thing is so
OOP'd to death and so-called COM'ed out, it's creating object for even the
simpliest of tasks.

Then, cause you got this object calling this (1)object and this (2)object
and this (3)object and this (4)object and this (5) object, you eventually
FORGOT what you wanted to do in the first place....HA HA HA....Like you
really know how to do Enterprise systems.......



Performance is not that easy after you done the coding. You still have to
RE-CODE again to fix all that crappy n-Tier architecture in the first place
SO in the end, you DON'T have n-Tier architecture....oh well, you just have
to learn the hard way.....






The business layer is a load of crap cause no one changes databases in the
first place for the same business functionality.......If you are going to
change the database, you are going to re-write the sprocs anyway on this
database port....you are basically designing in this modularity at the cost
of performance as well as reliability of the code in the first place.

The database modularity feature is a feature that has SO LITTLE BUSINESS
VALUE as it doesn't happen that often in the business world in the first
place.......I don't know of anyone who goes out and spends money on Oracle
and then switches to DB2 and throws out the Oracle License.

.....if anything, it's more cost efficient to buy faster hardware, keep the
license that you can't sell back to Oracle in the first place and install
the new server. That, in the end, is more cost efficient. It's called use
what you just purchased and can't return.

But let's say they switch to SQL Server from Oracle....now what are they
going to do with that Oracle DB then? They just paid tens of thousands for
it. Now you are saying, "Let's build a data-tier to be able to SWITCH BACK
from SQL Server to Oracle." That's not one, but TWO switches!!!

That's LUNACY...make up your mind!!!!


If you are going to make a switch, one of the things in deciding whether to
switch is going to be performance in the first place. And if it's
performance, why do you turn right back around and advocate a much less
performing n-Tier architecture????

At first, you say the reason to switch is due to performance(or size), so
you go from SQL to Oracle. Now you say it's

COST/PEFORMANCE, as you can get

two SQL Servers for the price of one Oracle. So you switch back!

The cost of the PORT is going to gobble up all the so- called cost savings of
this "so-called" modular data tier in the first place cause IT's NEVER that
easy..

And if your are big enterprise, they already have their code in place so
they are not going to need Mr. IT shops
modular-code-that-can-work-with-any-database, as they already picked a
database and have it running in production anyway.


Please stop with these silly arguments as I can easily use Microsoft OWN
marketing slogan against itself and I havent' even mentioned that one.......

See this threads....this guy lost big time.....

You as well as all MVP's, MCSD, Gurus have soooo much to learn.
tier+achitecture+for+Web%
3F+Why+group:microsoft.public.dotnet.*&hl=en&lr=&ie=UTF-
8&group=microsoft.public.dotnet.*&selm=%23lGPJeXZDHA.1640%
40TK2MSFTNGP10.phx.gbl&rnum=1


JUDGEMENT DAY FOR OOP COMPLEXITY Re: Reasons for a 3- tier ...

http://groups.google.com/groups?hl=en&lr=&ie=UTF- 8&threadm=enkCVqUfDHA.3284%
40tk2msftngp13.phx.gbl&rnum=3&prev=/groups%3Fhl%3Den%26lr%
3D%26ie%3DISO-8859-1%26q%3DReasons%2Bfor%2Ba%2B3-tier%
2Bachitecture%2Bfor%2BWeb%253F%2BWhy%26meta%3Dgroup%
253Dmicrosoft.public.dotnet.*


Hey eBay..Let's talk about OOP and

Reliability.........Re: Reasons for a

 

[Bob Lehmann]

You do realize that you are feeding a know-nothing troll, don't you?

Bob Lehmann

 

[David Sworder]

You do realize that you are feeding a know-nothing troll, don't you?

Bob Lehmann

This is true, but his post (Eddie's) was still interesting and
informative for me and others that are following this thread -- so it wasn't
wasted.

 

[Guest]

COMMENT INLINE BELOW

Dude,

Really... I've been working on N-Tier architecture for
several years now, as well as deploying that architecture
across various numbers of machines. (Sometimes all
deployed on one machine, other times deployed across
10+)...

What I have learned over the years is very simple.. Some
N-Tier projects fail simply because the developers that
are selected for the projects cannot understand simple
programming.

Whoa, wait a second!!!! "....developers cannot understand simple
programming"

If the medical profession said that some patients die because doctors don't
understand simple diagnosis we would be in a lot of trouble

How about the hardware engineers? If they said that some hardware fails
because they didn't understand the basics of electricity, their company
would fail immediately.

How about plain old construction workers? If they said what you said,
building and bridges might fail.

How about automotive workers and engineers? If they said what you said,
cars would not start in the morning....


What do all these profession have in common? They know the limitations of
their professions and the techniques they use. They use AUTOMATION when
possible...not some fancy stupid OOP model that's so complex it would take a
year just to learn the ins and outs.

Arrogant programmers think they every programmer should do it their way and
their way is the BEST WAY.

Well, since you are in denial about the failures of IT, let me wake you up!

One cannot say, "Oh, software failures are a part of life. And come around
and say that the current method is the best one we have." B.S.

Yes, I'll admin there is a tendency to "Over OOP" an
application, that can be an issue, but that normally
comes from inexperience... N-Tier and OOP are still
fairly new to many developers/architects, and as such is
miss used way to often...

Don't judge a programming methodolody on it's failure..
There are always failures.. Learn waht you can, try them
for yourself, and have a good look at how many Success
Stories there are.. You'll probably be very supprised...

WHOA!!!! wait ONE SECOND HERE.

"DON'T JUDGE A PROGRAMMING METHODOLODY ON IT'S FAILURE"!!!!!

Oh, OK.

Let's see.... if 2-Tier fails you will immediately point out that failure,
BUT when n-Tier fails, you are not allowed to point that out.

HA HA HA HA HA HA HA HA

You are REALLY FUNNY!!! Do you what a DOUBLE STANDARD is?

Let's see, the DOT BOMB FAILURES....OK, well since a lot of those were OOP
failures and we can't look at those, the DOT BOMB has now become DOT.COM
back from the dead????

But Cobol, of which is still being used to this day has had a stellar track
record...but of course we can't look at those success as the programming it
TOO simple to handle today's COMPLEX programming tasks..but when push comes
to shove in mission critical apps like Hospitals...they have proven
themselves...


OH BY the way, when a crappy OOP program kills someone, lawyers don't even
bother to look at EULA's

You say "What happens if a tier breaks.. " simpe.. Fix
it... What happens if you single layer application
breaks.. easy.. Fix...

And yes, project do quite often take > 1 year for
development... Imagine a government department, that
always has changing requirements... The project will
never finish.. that's why government departments
generally have very large IT departments.. Laws Change,
regulations change, that's something we have to deal
with... The N-Tier achitecture helps handle these
changes, without having to root through a single project
consisting of thousands and possibly millions of lines of
code...

If that is SOOOOO true, WHY, right NOW, is the IRS computer system so
screwed up? It takes forever to get your returns back......Why? right now
is the FBI system so screwed up there is a congressional investigation?

Congress is holding up money for the FBI's Trilogy modernization program
because of a projected 50 percent cost overrun.
http://www.fcw.com/fcw/articles/2003/0127/web-trilogy-01-28-03.asp


But I thought OOP was supposed to solve everything....2-second change,
n-Tier...blah blah blah......

Isn't OOP and n-Tier supposed to be GREAT for LARGE IT PROJECTS???

Doesn't seem so does it?

As for windows forms, what can you do for a client that
you can't with web.. simple... Try implementing a project
management system that allows dragging around resources,
try getting a user to select a single record form a large
set of data (It's not always possible to do with a drop
down combo ).. What about Imaging...

Yes, I am well aware of all the technologies that you can
use to do this sort of thing using a web based client,
but true usability is in windows apps...

Lets also talk development methodologies...

DEV/TEST/Release.. yep, it's old, so what.. it works...
To many companies ignore the testing, they release their
apps and guess what.. The app falls on it's arse,
management yells, and the project gets cancelled... then
someone mentions it was an N-Tier project... Next thing
you know, there is a case study on the internet about how
bad N-Tier is, all because management decided it was to
costly to do testing...
Testing is required, it's not an option.... TEST TEST
TEST...

NO, NO, NO, NO......
With OOP, you are still in the "Design the OOP Model" for at least a half
of year. Then, you have to code the stupid model.....and by that time, you
have ZERO budget to TEST........

You don't need to follow the strict DEV/TEST, Partial
testing can be done while still in development. Unit
testing can be done, and some usability testing can be
done.. It really doesn't have to cost alot...

Doesn't COST a LOT?

It will cost you the contract if you don't deliver it within a reasonable
time...

I think maybe you need to move around a little more, do
some contracting, get some more experience, and get out
of the little two bit company that you work for.. Because
obviously it hiding you from the real world..

Last but not least, don't believe everything you read on
the internet, especially if it's someone elses opinion on
slashdot

Tell that to the constant failures of IT....as we speak, I will always be
able to pull even more IT failures that make it to the news.....


You and the rest of the OOP fanactics need to get over your denial stage and
stop closing your eyes to the failures of IT.....or else your job can be
easily shipped to India as they can product crappy OOP software like you are
producing for a fraction of what you are charging......
TA TA!!!



Try not to embarrass yourself and do some soul searching....Just because the
MCSD test asked the question and you got it right doesn't mean it's going to
happen in the real world...OH that's right, MCSD are just a bunch of timed
questions...when it comes to actually doing the code and putting it into
production that's a different story as the test givers haven't a clue what
the real world is like cause they are so busy writing crappy simple demo
code that's never production........Hmmmm, I wonder what other .NET authors
and java author and C++ authors could be in the same boat.........sample
code and production code...big big difference...same for sample OOP and
production OOP (if you can ever get to production OOP)

(Note: again and again, Microsoft and .net authors are openly
challenged....yet do you see them here supporting you...nope)

 

[Guest]

COMMENTS INLINE BELOW......

I feel I have to respond to this rant. 18 months ? I should also tell
you that we've gone from 1 up to 15 developers working on this at any
one time, with a core of 10! This is because the business is
incredibly complicated. To give you an idea, the Oracle system alone
holds 3500 tables, and the business logic behind the AS400 system is
40 years old!

On the hardware side, by using 3-tier we've been able to site our web
servers externally, at an ISP, and have our app servers in-house, both
as part of web farms. If we need to scale out, this in now easy. The
business is time-critical. We can't just ask the customers to wait. We
just buy another server and add it to the cluster.


OK. Part of the system includes temps swiping in as they turn up for
work and leave. I don't believe you can do that through a browser. And
it has to work standalone, when the network fails.

9 to 10 people for an entire WINDOWS APP!!!!!
....just to punch their TIME CARD!!!!!!

You can buy a time card system at Office Depot for that and port it over as
a text file if need be.....

NINE to TEN PEOPLE!!!!


You know you can still have a web based system that logs them in and yes you
can have a card reader and you don't need a Windows app to do it.

Rich client??? Like what is it in that windows app that can't be
done (and

So you'd always build a web app no matter what ? Winforms does give
you a richer UI. And if we used dot net remoting, you could use it
away from the office. And if we use a smart client, deployment is easy
too. You should read the articles that were recommended in the
previous post.

RICHER UI doesn't mean RICHER profits... big difference....

If anything, it should be SIMPLIER UI...always SIMPLIER UI

Actually 10 boxes. But we could scale out the app farm or either of
the web farms.

2 external web servers,
external state server,
2 internal web servers,
2 app servers,
1 SQL Server,
1 AS400,
1 Oracle


And it's not finished yet! We've got another 2 years of work in the
pipeline.

Sounds like the Trilogy FBI project that's 50% over budget....

The business itself has just started to take off because we've put it
on the web!



That's why we've got 2 app servers. So if one fails, we can swap it
out without too many wories..

That's complete DUMB. I would always go for 4 web servers before a 2 + 2
setup that you have

I got proven Round Robin and/or load balancing tech out there that can do 4
web servers.

With your setup, you have have to have 2 levels of load balancing..which
means more programming, more complexity....


If one of the 4 goes down, I still have 75% capacity...but if, in your 2 + 2
Setup, if one goes down you have 50% capacity. PLUS, you won't easily know
where the problem is in the first place.

We could have redesigned the database around a new SQL Server, but
that really would be a large project.

I guess you presumed it was just me taking my sweet old time.

With 18 month, yes, you should have been built it around SQL Server....even
SAP can get their piece of crap going after a year or so....so with yours
being 1.5 years, you have no excuse....1.5 YEARS!!!!!

 

[Eddie de Bear]

hahahaha, this is funny...

Your analogies are funny... Doctors Miss Diagnose
problems all the time.. People don't always die becuase
of it...

Hardware is realeased with flaws all the time, that's why
more and more hardware have bioses that can be flash
upgraded... To fix problems that the hardware engineers
may have accidently introduced...

As for buildings.. we are talking about a profession that
has had a few thousand years to get it right... It's not
like it took them 35 years to go from nothing to building
bridges like the golden gate and buildings the the
pertonus towers... We are talking about a profession that
has standards, TESTING, TESTING and more TESTING before
something can be built.. Not to mention reviews, reviews
and more reviews during the build procedure..


Speaking about taking a year to learning the ins and
outs... Don't you find it strange that Doctors and
Engineers take several years to learn the ins and outs..
then they spend the rest of their life reading journals,
publications and attending forums to learn about the
LATEST way to do their job, more efficiently and
affectively.. They don't just keep giving out penicilan
and cutting off limbs with saws.. These professions
progress over time... They are also alot more mature...

Automotive engineers, once again, the basic internal
combustion engine has not really changed a great deal
since the first mass produced model T.. Sure, they have
moved the cams around, tweeked here, added electronic
fuel injection, but it still works the same.. So it's not
like they are designing completely different solutions
every time they start a new project...

Now, lets not actually take a look at the quotes you
plucked out of the air... I don't believe I ever said
that N-Tier was the only/best way.. It's just the way I
work...

Seen as you bought in COBOL, lets talk about that for a
second.. Yes, COBOL has been around for years, however
it's nowhere near as productive as as .NET, Java etc...
Wasn't that one of your main critisisms about the n-tier
model, it takes to long.. Sure, lets just use cobol and
every project will take twice as long... Sounds like a
good idea...

EULA's, where did you pluck that from??
Really, I think what you where trying to point out is
that poorly written software can kill people, correct...
Sure, that's why you need to implement good quality
control... Yes, I know it costs, but that should have
been factored into the cost of development.. Software
costs to develop, and if it is so sensitive that a
failure could cost lives, then I'm sure it's easy enough
to justify....

Design really doesn't take as long as you say.. Have you
heard of design patterns, re-usable code etc... Not to
mention that an application does not have to be fully
architected before development can start...

Testing, does it really cost a contract.. It does if you
submit a fixed price for the work that does not factor in
the testing.. Most clients I deal with are happy with
including the testing, as long as they see it.. Break
down the quote, into dev/test/deploy etc... People like
to know the software they get is going to be tested
(Helps with winning the contract...)..

Now lets look at what can happen if you deliver software
that has not been tested.. Your client sues you for all
your worth... (Not that you ever make a mistake when you
develop..)

Interesting.. Shipping off work to India.. I'm actually
from australia... And I have to say that so far very
little work has gone that way... So really, I'm not sure
I need to be worrying about that right now..

As for that well thought out shot about MCSDs.. I
actually produce a large amount of code.. Most of which
has made it's way into production.. (there is always a
small amount of code written, like TEST harnesses etc
that don't go to production..)

Don't take this the wrong way, but most other people on
this news group consider you a troll and not worth
arguing with, that's why they are not here arguing....

Me, I'm always looking for a better way to do things.. I
just need some real proof (Not MS/.NET/OOP bashing) that
what I'm doing is not the best for my clients.....

-----Original Message-----
COMMENT INLINE BELOW




Whoa, wait a second!!!! "....developers cannot understand simple
programming"

If the medical profession said that some patients die because doctors don't
understand simple diagnosis we would be in a lot of trouble

How about the hardware engineers? If they said that some hardware fails
because they didn't understand the basics of electricity, their company
would fail immediately.

How about plain old construction workers? If they said what you said,
building and bridges might fail.

How about automotive workers and engineers? If they said what you said,
cars would not start in the morning....


What do all these profession have in common? They know the limitations of
their professions and the techniques they use. They use AUTOMATION when
possible...not some fancy stupid OOP model that's so complex it would take a
year just to learn the ins and outs.

Arrogant programmers think they every programmer should do it their way and
their way is the BEST WAY.

Well, since you are in denial about the failures of IT, let me wake you up!

One cannot say, "Oh, software failures are a part of life. And come around
and say that the current method is the best one we have." B.S.


supprised...


WHOA!!!! wait ONE SECOND HERE.

"DON'T JUDGE A PROGRAMMING METHODOLODY ON IT'S FAILURE"!!!!!

Oh, OK.

Let's see.... if 2-Tier fails you will immediately point out that failure,
BUT when n-Tier fails, you are not allowed to point that out.

HA HA HA HA HA HA HA HA

You are REALLY FUNNY!!! Do you what a DOUBLE STANDARD is?

Let's see, the DOT BOMB FAILURES....OK, well since a lot of those were OOP
failures and we can't look at those, the DOT BOMB has now become DOT.COM
back from the dead????

But Cobol, of which is still being used to this day has had a stellar track
record...but of course we can't look at those success as the programming it
TOO simple to handle today's COMPLEX programming tasks..but when push comes
to shove in mission critical apps like Hospitals...they have proven
themselves...


OH BY the way, when a crappy OOP program kills someone, lawyers don't even
bother to look at EULA's






If that is SOOOOO true, WHY, right NOW, is the IRS computer system so
screwed up? It takes forever to get your returns back......Why? right now
is the FBI system so screwed up there is a congressional investigation?

Congress is holding up money for the FBI's Trilogy modernization program
because of a projected 50 percent cost overrun.
http://www.fcw.com/fcw/articles/2003/0127/web-trilogy-01- 28-03.asp


But I thought OOP was supposed to solve everything....2- second change,
n-Tier...blah blah blah......

Isn't OOP and n-Tier supposed to be GREAT for LARGE IT PROJECTS???

Doesn't seem so does it?











NO, NO, NO, NO......
With OOP, you are still in the "Design the OOP Model" for at least a half
of year. Then, you have to code the stupid

model.....and by that time, you

 

[Steve S]

The small mouse steps out of his house...... drags a cold FOSTERS
along.... and rolls it over to EDDIE....... "Have one on me Mate!"

Your very patient...... if he would only read what we are saying.... for
the value..... and not simply bash just to bash.... but then again.....
he has his opinion..... I just wish he would come out of the closet......
if he is ging to bash me.... the least he can do is have the guts to
identify himself...... other than NOSPAM......

The little mouse sniffs the air....... "OOPS..... dinner calls......."
and rushes back into his home...... ALOHA

 

[Steve S]

The little mouse slides his thumbnail back out oneto the floor...... this
time he wears a court robe and a white curly wig...... He clears his
throat.... takes a sip of water........ then announces.......

HEAR YE....... HEAR YE....... nospan..... has spoken...... and cited
another reference...... which has been pulled and inserted below (hey
careful where you insert that) :
*************
In the meantime, Trilogy's problems have been mounting. Last month, the
Justice Department's inspector general told lawmakers that the committee was
experiencing a cost overrun of $137.9 million for one year. When combined
with another overrun, it translated into a 50 percent increase in the
program's total cost.

"This is not a surprise," Senate appropriators said in a report filed
earlier this month. "The attempt to make up for 20 years of neglect in two
years of frenzied spending was destined to fail."

Although the committee has already provided a $100 million cushion in
supplemental funds, that money also has been used up.

"The FBI chose to squander this reserve. So when the funds are needed, none
are available," the report said.

Sen. Judd Gregg, the chairman of the Appropriations Committee panel that
handles the FBI budget, said in a recent speech on the Senate floor that
Trilogy has become a "disaster."

"Programs such as Trilogy do not need more money. What they need is more
management," Gregg said.

He said Congress "threw" too much money at the program last year to "show we
were concerned about terrorism." And as a result, he said, the program has
fallen apart.

"FBI software and hardware contracts for Trilogy have essentially become
gold-plated. The cost is soaring. The schedule is out of control," Gregg
said. "Right now we still do not have contracts on Trilogy hardware or
Trilogy software. We are completely at the mercy of the contractors

*************

Just for the record...... where in this..... did they say...... n-tier
or selected solution? I believe this is a result of poor project
management.... which is a totally different issue.....

Then again...... READ THE REFERENCES YOU ARE GOING TO USE...... TO PROVE
YOUR POINT...... it could save you from moments like these....

OH YEA..... if you want a piece of the action...... why don't you write
a proposal to the FBI..... that you recommend they do it all......
BROWSER BASED... ONE TIER, FIXED PRICE, NO TESTING AND IT WILL BE DONE IN
A FEW WEEKS!!!!!!! [the crowd gasps ...... and an immediate hush falls
over the masses] After all..... you keep bashing n-tier... and if you
truely are writing browser apps..... they are by defination...
MULTI-TIER..... simply due to how the webservers operate and run the
code...... BUT..... hey...... go for it.......

The little mouse steps off his thumbnail..... and slides it back to his
house.....

 

[Guest]

COMMENTS INLINE BELOW...

The little mouse slides his thumbnail back out oneto the floor...... this
time he wears a court robe and a white curly wig...... He clears his
throat.... takes a sip of water........ then announces.......

HEAR YE....... HEAR YE....... nospan..... has spoken...... and cited
another reference...... which has been pulled and inserted below (hey
careful where you insert that) :
*************
In the meantime, Trilogy's problems have been mounting. Last month, the
Justice Department's inspector general told lawmakers that the committee was
experiencing a cost overrun of $137.9 million for one year. When combined
with another overrun, it translated into a 50 percent increase in the
program's total cost.

"This is not a surprise," Senate appropriators said in a report filed
earlier this month. "The attempt to make up for 20 years of neglect in two
years of frenzied spending was destined to fail."

Although the committee has already provided a $100 million cushion in
supplemental funds, that money also has been used up.

"The FBI chose to squander this reserve. So when the funds are needed, none
are available," the report said.

Sen. Judd Gregg, the chairman of the Appropriations Committee panel that
handles the FBI budget, said in a recent speech on the Senate floor that
Trilogy has become a "disaster."

"Programs such as Trilogy do not need more money. What they need is more
management," Gregg said.

He said Congress "threw" too much money at the program last year to "show we
were concerned about terrorism." And as a result, he said, the program has
fallen apart.

"FBI software and hardware contracts for Trilogy have essentially become
gold-plated. The cost is soaring. The schedule is out of control," Gregg
said. "Right now we still do not have contracts on Trilogy hardware or
Trilogy software. We are completely at the mercy of the contractors

*************

Just for the record...... where in this..... did they say...... n-tier
or selected solution? I believe this is a result of poor project
management.... which is a totally different issue.....

OH NO IT"S NOT!

MANAGEMENT choose MCSD certified programmers and Java Certified porgrammers
and that oh so high FAILURE TRACK RECORD n-Tier Architecture.

Management does what the Programmers tell them to do. Management doesn't
know crap about actually doing the programming and you know it.

I challenge you to prove otherwise that management knows heads or tails of
what an IDE is or has even written a single line of code.. And those
management that did code at one time were really really crappy coders that
realized that they can survive with their people skills. These management
types have a hard enough time figuring out how to use MS Office let alone
know what VisualStudio.NET is.

THE BOTTOM LINE is that as a WHOLE, the PROJECT, and IT PROJECT in the
MILLIONS AND MILLIONS of DOLLARS FAILED....PERIOD....

IF n-Tier was SOOOOOO GREAT, it should have been able to at least overcome
management blunders like hiring MCSD's all over the place.

Then again...... READ THE REFERENCES YOU ARE GOING TO USE...... TO PROVE
YOUR POINT...... it could save you from moments like these....

SEE ABOVE for YOUR ANSWER


Why do you embarrass yourself?

OH YEA..... if you want a piece of the action...... why don't you write
a proposal to the FBI..... that you recommend they do it all......
BROWSER BASED... ONE TIER, FIXED PRICE, NO TESTING AND IT WILL BE DONE IN
A FEW WEEKS!!!!!!! [the crowd gasps ...... and an immediate hush falls
over the masses] After all..... you keep bashing n-tier... and if you
truely are writing browser apps..... they are by defination...
MULTI-TIER.....

UH OH, buzz word bingo redefining the term MULTI-TIER!!!

Since you have ZIPPO to go on, I will let you know that 2-Tier doesn't refer
to n-Tier...but as you want to twist these words out of proportion, go
ahead, you need to "think" that you had a chance of proving something that
you were right.

 

[Guest]

COMMENTS INLINE BELOW...

hahahaha, this is funny...

Your analogies are funny... Doctors Miss Diagnose
problems all the time.. People don't always die becuase
of it...

Yeah!....and guess what, they get sued!!!

Hardware is realeased with flaws all the time, that's why
more and more hardware have bioses that can be flash
upgraded... To fix problems that the hardware engineers
may have accidently introduced...

And they return it back to the store if the thing doesn't work.

As for buildings.. we are talking about a profession that
has had a few thousand years to get it right... It's not
like it took them 35 years to go from nothing to building
bridges like the golden gate and buildings the the
pertonus towers... We are talking about a profession that
has standards, TESTING, TESTING and more TESTING before
something can be built.. Not to mention reviews, reviews
and more reviews during the build procedure..

REVIEWS to admit their FAILURES......
Ah yes! Admit their FAILURES.......the FAILURES of n-TIER and OOP

Speaking about taking a year to learning the ins and
outs... Don't you find it strange that Doctors and
Engineers take several years to learn the ins and outs..
then they spend the rest of their life reading journals,
publications and attending forums to learn about the
LATEST way to do their job, more efficiently and
affectively.. They don't just keep giving out penicilan
and cutting off limbs with saws.. These professions
progress over time... They are also alot more mature...

YES, SPEAKING ABOUT Engineers and Doctors they ADMIT their MISTAKES.
Even the most widely USED techniques, medicines and building architectures
are CHALLENGED.

NOW, where are the CHALLENGES on n-TIER and OOP????
All of the sudden n-Tier came alone and no one has done and ROI on n-Tier OR
and ROI on OOP.

Oh, what's that?

How about an ROI on Web Services.....was it Gartner or Forrester who blew
the lid on that.

Automotive engineers, once again, the basic internal
combustion engine has not really changed a great deal
since the first mass produced model T.. Sure, they have
moved the cams around, tweeked here, added electronic
fuel injection, but it still works the same.. So it's not
like they are designing completely different solutions
every time they start a new project...

Now, lets not actually take a look at the quotes you
plucked out of the air... I don't believe I ever said
that N-Tier was the only/best way.. It's just the way I
work...

Seen as you bought in COBOL, lets talk about that for a
second.. Yes, COBOL has been around for years, however
it's nowhere near as productive as as .NET, Java etc...

If .NET and Java are so productive why are large projects in CRM failures?
How about the McDonald's IT project that was a failure.....


Seems like n-Tier and OOP are, at best, a HIT and MISS type of
architecture...win some and you lose some....

Wasn't that one of your main critisisms about the n-tier
model, it takes to long.. Sure, lets just use cobol and
every project will take twice as long... Sounds like a
good idea...

With the new IDE's now, COBOL will be very productive.....

EULA's, where did you pluck that from??
Really, I think what you where trying to point out is
that poorly written software can kill people, correct...
Sure, that's why you need to implement good quality
control... Yes, I know it costs, but that should have
been factored into the cost of development.. Software
costs to develop, and if it is so sensitive that a
failure could cost lives, then I'm sure it's easy enough
to justify....

Design really doesn't take as long as you say.. Have you
heard of design patterns, re-usable code etc...

GOOD! I am glad you mentioned DESIGN PATTERNS.

Here is how one can CRUSH the OOP ARROGANCE using their very own DESIGN
PATTERNS against it.

Now, let's see....Design Patterns come about from YEARS of EXPERIENCE
right?

It says design your object model using these types of patterns for these
types of situations.

AH, BUT OOP and n-TIER is supposed to be FLEXIBLE and EXTENSIBLE and is
easily able to handle *CHANGE*.

BUT WAIT a SECOND, didn't *Design patterns* say that if your don't pick the
right pattern, you WILL REALLY SCREW up the entire thing.......

CAN YOU SEE THE COMPLETE CONTRADICTION???????????

Ah YES...the GANG OF FOUR.......

HA HA HA HA HA HA HA HA!!!!!!!

How many top notch programmers pick up a book and automatically assume it
etched in stone?


Most BOOK authors have NEVER WRITTEN PRODUCTION CODE....they sit around in
the design phase and in the nice offices telling other programmers what to
do.....

They have never written an ENTERPRISE system all by themselves and maintain
it by themselves.......they just tell other programmers to follow their
object model and UML diagrams and fix their mistakes........BIG
DIFFERENCE!!!!!

Kinda reminds me of the Chief Engineers who design cars but forget that some
factory worker has to put the thing together.

Ah....and the same goes for those MCSD test authors.....



AHHHHH,.....But there is MORE!!!

JUST LIKE the PREVIOUS thread and the previous person, each and every time,
I can bring up a TOTALLY NEW POINT he and you have NEVER THOUGH OF.....

POST after POST....something new...something that by itself can tear down
the OOP HIGH MINDEDNESS and n-Tier architecture.

That person has never attempted to answer the RIDDLE.

FULLY OOP = _____________ = FULL PLATE OF PROBLEMS



OHH, there is another BOOK...standard Microsoft reading, "CODE
COMPLETE"......well, I can already see the results of that book's influence
daily so I will reserve my point until later........Ha ha ha ha ha!!!!!


Now, WHO DO YOU think you deal with?

No Microsoft employee here to back you up....if there was a reasonable
argument they could make, they would be here...but no, they are no where to
be found.......

Not to
mention that an application does not have to be fully
architected before development can start...

Testing, does it really cost a contract.. It does if you
submit a fixed price for the work that does not factor in
the testing.. Most clients I deal with are happy with
including the testing, as long as they see it.. Break
down the quote, into dev/test/deploy etc... People like
to know the software they get is going to be tested
(Helps with winning the contract...)..

Now lets look at what can happen if you deliver software
that has not been tested.. Your client sues you for all
your worth... (Not that you ever make a mistake when you
develop..)

YEAH!!! so what's your point?
They, the n-Tier and CRM and McDonalds and the Trilogy have failed.

And some, mainly simple CRM's are being SUED right this minute.

Interesting.. Shipping off work to India.. I'm actually
from australia... And I have to say that so far very
little work has gone that way... So really, I'm not sure
I need to be worrying about that right now..

As for that well thought out shot about MCSDs.. I
actually produce a large amount of code.. Most of which
has made it's way into production.. (there is always a
small amount of code written, like TEST harnesses etc
that don't go to production..)

Don't take this the wrong way, but most other people on
this news group consider you a troll and not worth
arguing with, that's why they are not here arguing....

TROLL or the TRUTH TELLER?

They don't argue cause they don't have the truth.......


Can you argue against the truth? no you can't. that's why they and you can't
argue.

Me, I'm always looking for a better way to do things.. I
just need some real proof (Not MS/.NET/OOP bashing) that
what I'm doing is not the best for my clients.....

It's because your clients don't know any better and don't expect much from
I.T. as I.T. fails them all the time....

 

[David Noble]

You've clearly never worked on a large project. The government could
pass new laws, users could demand new features, and we'd have to
re-work what we've done. It's the nature of the business we're in. Why
do you think they're still developing apps like SAP, Excel or Oracle
Apps ?