On Sun, 1 Feb 1998 22:12:27 -0500, Nick Demos wrote,
[snip...]
>> Which caused an industry wide response...
>>
>> 1. Standard, more open interfaces to data sources.
>> 2. Host based Internet/Intranet computing services.
>> 3. Give users productivity gains over work process.
>> 4. Enterprise wide services.
>> 5. Multi-tier architectures.
>OK on 1 through 4, but 5 is an ERROR. WE need a thick (or
>sometimes thin) client a communications network
>and servers. A third layer just adds unnecesary complexity.
>Put another way, one can have as many layers as one wants, but
>as long as one has server to server communications capability,
>one only needs two kinds of equipment and perating systems -
>servers and clients.
I believe this to be incorrect, Nick. N-tier support is a better expression
of concept than simple client-server. Keeping within the original analogy,
maybe we can express N-Tier situations as similar to the number of wheels on
a vehicle. In some cases, 2 wheels is the best choice, while in others 4
wheels is the most common. Kids tricycles and motor bikes for parapalegics
are best served by 3. And of course to really haul mass, 18 wheelers are
tops.
N-tier, for any value greater than 2 suggests that at least one of the tiers
is both client AND server. These are the tiers that handle the business
rules. Moreover, I would suggest that while there are some cases of
applications where clients are just clients and servers are just servers,
there are many more applications that are better handled as 3 or more tiers.
But as I said before, the number of tier levels is an expression of concept,
not necessarily a distribution plan - you can still write 3-tier yet run it
on a single machine. So tiers doesn't fix the number of operating systems,
or equipment types.
IMHO.
|----
|Jim Wowchuk Vanguard Computer Services
| _--_|\ Email: [log in to unmask]
|/ \ Post: PO Box 18, North Ryde, NSW 2113
|\_.--._/ <---Sydney NSW Phone: +61 (2) 9888-9688
| v Australia Fax: +61 (2) 9888-3056
|