TCLUG Archive
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [TCLUG:12613] Reasons not to use Apache?



On Thu, 20 Jan 2000, Dave Sherohman wrote:

> A previous programmer planted the idea in my boss's head that writing an
> Apache module to do this would be a good and readily portable way of handling
> it.  I disagree and think that it would be much better to create a separate
> standalone server to handle these functions.  Philosophically, I believe that
> this would be the way to go because I favor the *nix 'lots of little pieces
> that each do one thing very well and can be readily combined with each other
> to handle bigger tasks' approach.

Sure. But that philosophy is one for batch-job software, not client-server
software. What peer software will your server have? Is it going to pass
batches of data along to other custom servers you write?

> Practically, running this through a general-purpose web server and HTTP
> connections is sure to add a _lot_ of overhead.  Additionally, the
> management functions are much better suited to a connection-based
> protocol than a connectionless one.

You'd have to prove that to me. Can you post your docs?

> Aside from overhead issues (bosses are not widely known for being swayed by
> philosophy),

Can you blame them?

> what other reasons can I give for not doing this in Apache?

The only compelling one that I can see would be the
connction/connectionless thing, maybe.

> The two reasons he's mentioned for favoring Apache that I can't readily
> counter are availablilty of load-balancing options other than
> round-robin DNS entries (I'm sure they're out there, but I don't know of
> any) and a desire to minimize the number of processes running on each
> machine (which is, essentially, a philosophical matter).

Minimizing discrete server processes is a practical matter, if you ask me.


--
Christopher Reid Palmer : www.innerfireworks.com

Let go of the hangers-on.