LAMP & Ubuntu

This should be easier, but I think made harder because I corrupted my Firefox applications helper list, and then chose the wrong way out

My problem, the browser tries to download index.phtml, in retrospect because, I believe that the module wasn’t properly installed into apache. Using Ubuntu 9.04

The Right Way

This page, Apache MySQL PHP at https://help.ubuntu.com/ is the bible.

apt-get install  apache2 php5 php5-mysql libapache2-mod-php5 php5-mysql mysql-server

I am running in a virtual box so I needed to create a /etc/apache2/conf.d/fqdn. There wasn’t one on my system, it needs the line

ServerName localhost

I suspect that fixed it, and that my errors were caused by not initially correctly installing libapache2-mod-php5. I have proved that my install works by using Opera. Now to remove the .phtml line from the firefox browser and make the vm a server.

Short URL’s

Microblogging makes this important. There are a bunch, but I prefer http://is.gd because its five characters long.

Services

Other’s I’ve found and used include

A number of people publish top 10 lists, you can use google to find them.

Problem definition

I am interested in writing or implementing a url-shortener, for use inside the firewall, or implementing through an API. It seems quite simple until you consider performance at scale. My first thought was to use a database with an identifier/sequence data type as the key to the hashing algorithm, since its best to keep a record of the URL’s issued so if people ask twice for the same url it can be reused.

The second part is resolving the URL on request. Is this just an apache solution with a very large redirect file?

Some research

I used google: url shortner howto, to see what it came up with.

Two interesting finds included

Ideas

Idea No 1 was to use a database identifier type and hash it, I was going to investigate if we could use a web server redirection file, or if I’d have to write something using a db retrieval. However both these the database transactions introduce serial bottlenecks and inhibit scalability.

So can pre-allocate the sh.urls in blocks , and allocate them using hadoop? Is hadoop a sensible solution for “find me next unused”, or perhaps it doesn’t matter; its just find me one unused slot?