Is the LAMP stack appropriate for Enterprise use?

Question

Is the LAMP (Linux, Apache, MySQL, PHP / Ruby / Python) stack appropriate for Enterprise use?

To be clear, by "Enterprise", I mean a large or very large company, where security, robustness, availability of skill sets, Total Cost of Ownership (TCO), scalability, and availability of tools are key considerations. Said another way, a company that looks for external adoption of frameworks / architecture - Something ubiquitous will be seen as more "valid" than something exotic / esoteric in this kind of environment.

I've seen use cases where Oracle, IBM, and Sun have implemented systems on the LAMP stack for various Enterprises. I've also seen examples where websites like yellowpages.com (Ruby on rails) and Facebook (php) are built on it. However, none of these examples are exactly what I'm looking for.

I'm really trying to find examples where it is an Enterprise standard at a very large bank (I.e., Citigroup), Telecom company (I.e., AT&T), or manufacturer (I.e., Proctor and Gamble). Just to be clear, I'm not looking for an example where it's used in a limited sense (Like at JPMorgan Chase), but where it's a core platform for systems like CRM, manufacturing systems, or HR management, as well as for internal and external websites.

The perception I've seen so far is that applications built on the LAMP stack perform slower and are less flexible. Some of the arguments I've heard are:

• Linux is seen as not as well supported as Unix, Solaris, or Windows Servers.

• Apache is harder to configure and maintain than web servers like BEA WebLogic or IIS.

• MySQL is a "not ready for prime time" DB for hobbyists, and not a competitor for SQL Server or Oracle (Although PostgreSQL seems to have a reputation for being more robust).

• PHP / Ruby on rails are optimized for CRUD (Create, Read, Update and Delete operations). Although this is an advantage when building CRUD-intensive web aplications, both perform slower than Java/Java EE or C# (which are both common Enterprise standards). Furthermore, a lot of applications and systems (like manufacturing systems) have a lot of non-CRUD functionality that may be harder to build with PHP or Ruby, or even Python.

Can anyone please provide arguments to support or refute the idea of the LAMP stack being appropriate for the Enterprise?

Thanks!

KA

UPDATE: Some times the LAMP Stack is Appropriate for Enterprise Use: Externally-Facing Blogs

Answer 1

"but where it's a core platform for systems like CRM and HR, as well as for internal and external websites"

First, find a LAMP CRM or HR application.

Then find a customer for the LAMP CRM or HR application.

Sadly, there aren't a lot of examples of item 1. Therefore, your case is proven. It can't be used for enterprise applications because -- currently -- there aren't any of the applications you call "enterprise".

Your other points, however, are very interesting.

1. Linux is seen as not as well supported as Unix, Solaris, or Windows Servers. I think Red Hat would object strongly to this. Give them a call. I think they'll make a very persuasive sales pitch. Read their success stories.

2. Apache is harder to configure and maintain than web servers like BEA WebLogic or IIS. By whom? Apache web site managers? Or IIS web site managers? This is entirely subjective.

3. MySQL is a "not ready for prime time" DB. Take it up with Sun Microsystems. I think they'd object strongly to this. Give them a call. I think they'll make a very persuasive sales pitch. Read their success stories.

4. PHP / Ruby on rails are optimized for CRUD, and both are slowly performing. Could be true. Java and Python might be faster. PHP and Ruby aren't the last word in LAMP.

Answer 2

Something ubiquitous will be seen as more "valid" than something exotic / esoteric in this kind of environment.

Although I personally wouldn't recommend PHP due to the many flaws in the language, it's most certainly ubiquitous. With the advent of phusion passenger, Rails support amongst shared-hosting companies is growing pretty quickly too. I give it another year or 2 at most before 90+% of shared-hosting accounts support rails out of the box. If that's not ubiquitous, what is?

Linux is seen as not as well supported as Unix, Solaris, or Windows Servers.

If this bothers you, purchase support from RedHat, or install Solaris and purchase support from Sun. Both of those will give you just as good support as Microsoft is likely to

Apache is harder to configure and maintain than web servers like BEA WebLogic or IIS.

I can't speak for BEA WebLogic, but having configured both Apache, IIS, and Tomcat, Apache is the easiest both to understand, and to find examples and documentation for by a long way.

MySQL is a "not ready for prime time" DB for hobbyists, and not a competitor for SQL Server or Oracle.

Oh really?. You should make it your mission to tell NASA, Google, CERN, Reuters etc that they're all using a hobbyist database that isn't ready for prime-time.

PHP / Ruby on rails are optimized for CRUD, and both perform slower than Java/Java EE or C# (which are both common Enterprise standards).

There are 2 things here:

Optimized for CRUD - This is totally irrelevant.
Rails and some of the python/php frameworks are optimized for CRUD apps. Many of the C#/Java frameworks are also optimized for CRUD apps. However, if the app you're building is a CRUD app (and 99% of web applications are), isn't this a Good Thing?
If you're not building a CRUD app, there are plenty of non-crud-optimized frameworks in ruby/python/php/java/C#. Net win: Nobody (hence it's irrelevant)

Perform slower than Java/C# - This is undoubtedly true, but it also doesn't matter. For a low-traffic site the performance difference isn't going to amount to anything, and for a high-traffic site your bottleneck will be the database, whether it be MySQL, oracle, or whatever.

What you trade-off for all of this is development time. Once you've used all this advice to convince your boss that you won't lose out on anything by using LAMP, If you crunch the numbers and show your them that it is going to take 6 man-months to build the site in Java, and only 3 to build it in ruby/python then that's really what it comes down to.