ASP, CGI and PHP Scripts and Record-Locking: What Every Webmaster Needs To Know   by Sunil Tanna

Many of us install server-side (ASP, CGI or PHP) scripts on our


web sites, and many of this scripts store data on the server.


However, poorly designed scripts can experience performance


problems and sometimes even data corruption on busy (and not so


busy) web sites.

If you`re not a programmer, why should this matter to you?

Answer: Even if you`re just installing and using server-side


scripts, you`ll want to make sure that the scripts that you


choose don`t randomly break or corrupt your data.

First, some examples of the types of scripts which store data on


web servers include:

(Of course, many scripts in each of these (and other) categories


are well-designed, and run perfectly well even on very busy web


sites).

1. Follow-up autoresponders typically store the list of


subscribers to the autoresponder, as well where in the sequence


of messages, each subscriber is. Examples of autoresponder


scripts: http://www.scriptcavern.com/scr_email_auto.php

2. Classified ad scripts store (at least) a list of all the


classified ads placed by visitors. Examples of this type of


script: http://www.scriptcavern.com/scr_classified.php

3. Free for all links scripts store a list of all links posted by


visitors. See some example scripts listed at:


http://www.scriptcavern.com/scr_ffa.php

4. Top site scripts usually store a list of the members of the


top site as well as information about the number of `votes` that


each has received. For examples of this type of script, see


http://www.scriptcavern.com/scr_topsite.php

So what kind of scripts have problems? And what sort of problems


am I talking about?

Well the principle problems all relate to what happens when bits


of data from multiple users needs to be stored on updated at the


same time. Some scripts handle these situations well, but others


don`t...

DATA CORRUPTION

Here`s a common data corruption problem that can occur with many


scripts:

1. When some bit of data needs to be updated, a copy of the


server-side script starts running, and then starts updating it.

2. If another user comes along and does an update before the


first copy of the script has finished, a second copy of the


script starts running at the same time.

3. There are a number of ways things can now go wrong, for


example:

(a) What if the first copy of the script reads in the data, then


the second copy reads the same data, then the first copy updates


the data, then the second copy updates the data? Answer: any


changes made by the first copy of the script can get lost.

(b) What if the first and second copy of scripts are both adding


multiple bits of new data to the store at the same time? For


example, imagine each needs to store the headline, description


and the name of the person posting a classified ad. Well, what


can happen (with some scripts) is the two classified ads can get


intermingled, so you might get (for example) HEADLINE-1,


DESCRIPTION-1, HEADLINE-2, PERSON-1, DESCRIPTION-2, PERSON-2.


Or worse yet, you might get bits of each part of each classified


ad, mixed with the bits of the other. This type of thing is


usually really bad news, as your data may consequently becoming


unusable from that point on.

Does this sound too unlikely a problem to worry about? Don`t bank


on it... even if it happens only 1 time in 1,000, or 1 in 10,000,


eventually it will happen: You need a solution.

So the real question is: is it possible for programmers to create


scripts without these kinds of problems? Fortunately the answer


is yes, and there are a number of ways that programmers can


address it:

1. They can store each bit of data in a separate file. This isn`t


necessarily a total solution by itself (in particular, a script


which just does this could still have problems if multiple copies


of a script update the same file at the same time), but it does


make data corruption less likely, and if corruption does occur,


at least it won`t corrupt the entire data store in one go.

2. They can use file-locking. This means that if one copy of a


script is working with a file, another copy of the script is


prevented from working on that file, until the first copy has


finished. File-locking works if done correctly, but programming


it into a script needs to be done very carefully and precisely,


for every single possible case... even a tiny bug or omission can


allow the possibility of data-corruption in through the backdoor!

3. They can use a database (such as MySQL) to store the data.


Provided the data is properly structured in the database, the


database handles the locking automatically. And, as the


programmer doesn`t have to write their own special locking


routines, the possibility of bugs and omissions are much reduced.

PERFORMANCE PROBLEMS

Of course, avoiding having your data corrupted should be the


paramount consideration in choosing a script, but is there


anything else we need to be concerned about?

Answer: Performance

Of course, all webmasters are aiming to build busy high traffic


web sites... but will your scripts be able to handle the load?

Go back and re-read the paragraph on file-locking. Now think about


what would happen if all the classified ads on your classified page


were stored in a single file (or all the links on your top site, or


all the subscribers to your autoresponder, etc.).

What would happen?

Answer: Because each update can only be performed after the previous


update has been completely finished, your site may be slow, or even


unable to handle all your users` requests.

So what`s the solution?

There`s two options that programmers can use:

1. They can use lots of small files and file-lock each


individually (for example, one per classified, one per top site


listing, etc.). Of course, this needs to be handled very


carefully...

2. They can use a database (like MySQL), as databases allow any


one individual record (`row`) to be updated, even when another


is also being updated.

IN CONCLUSION

Now, let`s summarise:

1. Scripts that store data in files need to use file-locking


to avoid data-corruption, and they also need to break the data


into separately updateable chunks to avoid performance problems


on busy web sites.

2. Scripts that store data in databases (like MySQL), provided


of course that they have been properly coded, are usually less


likely to suffer from data-corruption or performance problems.

And one additional point:

3. Even the best script is not immune to hard-disk hardware


failures, your web host being struck by lightning, and all the


other snafus that can happen. So, do take regular back-ups of any


data that you can`t afford to lose!

In short, even if you`re not a script programmer, you need to be


aware of data storage issues. In future, when considering a


script for your web site, don`t be afraid to ask some hard


questions about how it stores data and how well it handles multiple


users.

This article is Copyright (C) 2005, Answers 2000 Limited.

About the Author:


This article was written by Sunil Tanna of Answers 2000. For a


directory of ASP, CGI, PHP and Remotely hosted scripts, please


visit http://www.scriptcavern.com - and for scripts written by


Answers 2000 please visit http://www.scriptrocket.com


-----------------------------------------------------------------


-----------------------------------------------------------------

Publication Terms And Conditions:

Answers 2000 Limited grants you a free non-exclusive permission


(license) to publish a copy of this article on your web site or


opt-in ezine, subject to you complying with ALL of the following:

1. You must publish the article in full and unedited (except that


you may omit this Terms and Conditions section, you may omit the


word count, and you may correct any typos that you might find).


2. If you publish on a web site: (i) you must make ALL links


clickable, (ii) you may format the article to fit within your web


site`s design, (iii) you must include the copyright notice and


`About the Author` section at the end.


3. If you publish in an ezine: (i) your ezine must be opt-in with


your users having specifically elected to subscribe to your ezine


and with the ability to unsubscribe at any time, (ii) you must


include all link URLs unedited and in full, (iii) you may format


the article to your ezine`s layout, (iv) you must include the


copyright notice and `About the Author` section at the end.


4. To the maximum extent permissible under law, this article is


provided `AS IS` without warranties of any kind whether express


or implied.


5. These terms and conditions shall be governed by and construed


in accordance with the laws of England and Wales. Any disputes


arising from matters relating to this article shall be


exclusively subject to the jurisdiction of the courts of England


and Wales. You agree that any legal action against Answers 2000


Limited (or its directors, officers, or employees) relating to


this article or this agreement will be brought in the courts of


London, England, however Answers 2000 Limited reserves right to


pursue breach of these terms in any jurisdiction.

There are 1225 words in this article (including title and About


the Author section).

-----------------------------------------------------------------