Bulman Development Ltd

Bulman Development Limited is my company. Most of what I do for consulting and contracting goes through it for invoicing, and tax reasons.  My standard consulting rates are £500 per day, or £100 per hour.

Areas of Expertise:

• PHP Scaling – See phpscaling.com
• Beanstalkd & other Job queuing – Beanstalkd experts on Stack Overflow.com
• LAMP system optimisation – Linux OS, Apache, Mysql and PHP code optimisations

Some examples:

⇒ At Binweevils, I joined knowing that they were planning to advertise the site on multiple childrens TV channels in summer of 2010. Thy had previously done so in Summer 2009 – but the website did not survive that.  I spent the first few months re-building the infrastructure with version control, an internal development server and optimisations to the database, web-server & main code-paths, and I’m very happy to report that very few problems occurred when the TV adverts rolled that summer, along with numerous other new features, large and small that were rolled out along the way. 

We rounded out the year with over 10,000 concurrent players on the website in the run-up to Christmas 2010. BinWeevils went on to win the BAFTA Kid’s website awards in 2011, 2012, 2013 & 2014 (taking the crown from Club Penguin, the winner in 2009, & 2010).

It’s odd to have fan-pages written about you….

⇒ Just before Christmas 2012, I was asked to help speed up a process that pre-generated content from a Mysql server. The task was taking around thirty minutes each time. I used my skills to examine and optimise the MySQL server, and now it takes five minutes or less per run. This has enabled updates to the website to be run far faster, and more often.

⇒ I was given access to around 85 million twitter profiles screen names. The task was to query the Twitter API for each one, and pull information for some further processing before putting them into the database for use.

I wrote a system to put 100,000 at a time into beanstalkd, keeping the queue topped up as required, and then fetch 100 at a time from the queue, query the Twitter API and put the results into a database. Within a matter of days, it was fetching over 3.5 million profiles per day with 3 concurrent workers. If the database write-speed was not a limiting factor, then ten times that amount would have been trivially possible.