Igor Sysoev started the development of nginx in 2005. It’s now known as a high performance webserver with a very low memory footprint. Already 4% of all webservers are running nginx according to a research done by Google in 2007, well-known sites like Wordpress.com and Hulu are some good showcases.

Some of my servers are now running nginx for about a week and the Russian software actually outperformed my expectations!  The load average decreased with 70% and I now have plenty of RAM available during peak-times. I did have to invest quite some time to migrate from Apache, but it saved me a lot of $$$ on buying extra hardware. I believe anyone operating a high traffic website with both static and dynamic content could benefit from a migration to nginx.  To make the migration somewhat easier, here’s a list of resources I used during the migration:

- Howto install nginx with PHP5:
http://www.cyberciti.biz/faq/rhel-fedora-install-configure-nginx-php5/

- Convert .htaccess rewrite rules to nginx rewrite:
http://blog.taragana.com/index.php/archive/nginx-hacking-tips/

- Set up loadbalancing with nginx:
http://wiki.nginx.org/NginxLoadBalanceExample

- Nginx wiki:
http://wiki.nginx.org/Main

My expectation is that we’ll see nginx showing up more and more serving high traffic websites. It’s a very good alternative for Apache, offering comparable functionality with a huge performance improvement.

]]> http://www.jochemprins.com/2009/09/20/recommended-migrate-from-apache-to-nginx/feed/ 0 http://www.jochemprins.com/2009/04/27/book-review-maverick-by-ricardo-semler/ http://www.jochemprins.com/2009/04/27/book-review-maverick-by-ricardo-semler/#comments Mon, 27 Apr 2009 17:39:18 +0000 Jochem http://www.jochemprins.com/?p=161

There’s a reason for this being the first book review posted on my blog: this book should definitely be on the reading-list of every entrepreneur or manager. I think Ricardo Semler is a true visionary regarding organizational structure and I believe every company could learn from his practices.

In Maverick, Ricardo Semler tells the story of how he transformed the company he inherited from his father in the 1980’s. The Brazilian manufacturing company, Semco, had a (traditional) hierarchical structure when Ricardo took over the business from his father. The book describes how Ricardo started changing the organization into a circular structure where every employee knows what’s going on in the company and, more importantly, understands what their own role, responsibility and value is.

The following excerpt from the book describes the effect of bureaucracy in a company, especially with respect to innovation. From my own experience I can only say that I fully agree:

In their quest for law, order, stability, and predictability, corporations make rules for every conceivable contingency. Policy manuals are created with the idea that, if a company puts everything in writing, management will be more rational and objective. Standardizing methods and conduct will guide new employees and insure that the entire company has a single, cohesive image. And so it became accepted that large organizations could not function without hundreds or thousands or tens of thousands of rules.

Sounds sensible, right? And it works fine for an army or a prison system. But not, I believe, for a business. And certainly not for a business that wants people to think, innovate, and act as human beings whenever possible. All those rules cause employees to forget that a company needs to be creative and adaptive to survive. Rules slow it down…

With few exceptions, rules and regulations only serve to:
1. Divert attention from the company’s objective.
2. Provide a false sense of security for the executives.
3. Create work for bean counters.
4. Teach men to stone dinosaurs and start fire with sticks.

The desire for rules and the need for innovation are, I believe, incompatible…Rules freeze companies inside a glacier; innovation lets them ride sleighs over it…A turtle may live for hundreds of years because it is well protected by its shell, but it only moves forward when it sticks out its head.

I honestly believe that employees who are not bothered by corporate rules, policies and regulations will always be more productive, creative and motivated. Trust the people who are working for your company, give them responsibility and they will take the responsibility. Don’t know where to start? Here are some of Ricardo’s best-practices:

• Make each business unit small enough so that those involved understand everything that is going on and can influence the outcomes.

• Share all information and eliminate secrets. You can’t expect involvement to flourish without an abundance of information available to all employees.

• Allow employees to set their own salary. Consider these criteria: what they think they can make elsewhere; what others with similar skills and responsibilities make in the company; what friends with similar backgrounds make; how much they need to live on.

• Demonstrate trust by eliminating symbols of corporate oppression as well as the perks of status.

This is only a small part of all the ideas and practices you’ll come across while reading Maverick. I can definitely recommend it, provided that you’re not afraid to change your thoughts about how successful business should be organized.

The first version is now ready (although there might be some bugs here and there), it’s available on www.tweepz.com.

Besides performing a regular search, you can also use a couple of operators in order to restrict the search to a specific field (bio:, name: and loc:). Next to these operators, you can perform fuzzy queries using spellslike: and soundslike: or by using a wildcard *. Think I’ll add an advanced search window later this week, that should make it a bit easier to perform advanced queries.

Currently around 400.000 twitter profiles have been indexed so there’re still a lot to collect by the crawler. If you are twittering, you can speed up the indexation of your own profile by following @tweepz.

Hope you’ll find it a handy tool, suggestions or feature requests are of course welcome!

The main problem researchers are facing is that each person’s style of speech is very different. And, especially if more people are in the same conversation, the speech recognition technology should be able to deal with all those different styles and vocabularies. I don’t expect it to take many more years before technology will be able to deal with those complications but there is some good news already! 

For some (actually quite a lot) possible applications of speech recognition, you don’t require a 100% accuracy. If it’s not strictly necessary to provide a completely accurate transcription of the speech, the current technology can already be very useful. For example, wouldn’t it be nice to:

- Automatically transcribe your phone calls, meetings, and personal memo’s into text for later reference
- Search through the content of a podcast archive for interesting episodes
- Search through the content of video’s for interesting shows

In order to achieve the latter, Exalead has partnered with LIMSI and developed Voxalead. With Voxalead, you can search through the content of the video’s and directly play the video with the embedded player. The player displays the transcription of text next to the video and clicking on a word brings the video directly to this precise time.

There will definitely be plenty of errors in the transcription, but… it doesn’t matter! The goal is to search through video’s and if some words are not recognized, no problem at all. I think it’s an excellent example of how to utilize the current speech-to-text technology in a useful way. What do you think?

Within a couple of minutes I already found a couple of php-scripts that perform the trick. After reading through the code and checking some comments on blogs I ended up with two candidates:

1) A script by Andy Moore on mobiForge
2) A script by Russell Beattie

That left me with the next dillema, which one to pick?

In order to answer this question, I defined the 2 main criteria that are important for the implementation on my website:

1) Code execution time: on a high traffic website, you want the code to be as efficient as possible
2) Mobile device coverage: detect as many mobile devices as possible (coverage)

This is probably a good time to mention the great WURFL project (open-source). WURFL is absolutely a great resource when you’re developing for mobile devices. Essentially, it’s a XML configuration file which contains information about capabilities and features of many mobile devices. I used this resource for testing the scripts above.

Testing and results

For both candidates, I looped through all user-agents in the WURFL file (9457) and tested:
- the average time required to check if this user-agent belongs to  mobile device or not
- if this user-agent is detected as a mobile device or not

And the results are…

Average execution time
- mobiForge script: 0,0078 seconds
- Russell’s script: 0,0021 seconds

Device coverage
- mobiForge scripts: 82,8%
- Russell’s scripts: 81,9%

In other words, Russell’s script runs faster but the script from mobiForge has a slightly better coverage. However, I think the small difference in coverage can almost be neglected since the biggest part of the mobile web users are all using one of the popular mobile devices (N95, HTCs etc). Both scripts have absolutely no problem detecting these. Furthermore, you can easily extend the list of mobile user-agents in both scripts (I already added the iPhone). Based on these findings, I decided to implement Russell’s script and it’s working just fine.

Oh… and in case you’re wondering, the mobile version of vertalen.nu will be released within the coming week.

I really recommend to have a look at Exalead Cloudview which, in my humble opinion, defines how the future of enterprise search will look like. And, while you’re at it, give Exalead’s Desktop search a try. The new desktop version can be accessed using your webbrowser and provides you with the same handy refinement options as can be used on exalead.com.

It’s very easy to setup, this is what I’ve done:

1) In Gmail, go to settings -> filters -> create a new filter
2) In the subject field, enter “todo:” -> click on ‘Next Step’
3) Check ‘Skip the inbox’
4) Check ‘Apply label’ -> new label -> enter “Todo” as label name
5) Click ‘Create Filter’ and your done!

Now write yourself an email with a subject like “todo: put out the garbage” and you’ll notice that Gmail will automatically list the mail in your todo list. You can do the same for a shoppinglist or readinglist, just use another prefix and label when you create the filter (e.g. “to read:”). Once you have finished the task you can either mark it as read (for your archive) of remove it completely.

You can now easily manage your todo’s anywhere & anytime using a webbrowser, email client or your mobile phone. For example, I found it really handy to be able to access my shoppinglist at the moment I actually need it.

At a career management course for HBS alumni, I learned that a person typically gets sick of a job after 18 months. This is a natural cycle, as you go through the excitement of learning a new job, become expert at it, and then gradually, it gets routine. So the advice I got was to plan for job obsolescence every 18 months. This didn’t mean that I had to leave the company and go to a new place – it had more to do with redefining my current job first to incorporate new challenges.

I reflected this on my own situation and came to the conclusion that I indeed arrived in the phase were a lot of things become a routine. For my job, it’s crucial to stay creative and come up with new innovations so routine is often a bad thing.

So… what’s my next challenge? Together with two new collegues, we’re going to start the dutch/benelux office of Exalead. Exalead is one of the major players in the enterprise search market. I know exalead pretty well since they are providing search technology to ilse since 2006. I’m a big fan of their technology and I’m very happy to become part of the company!

What we need to do is open up the social web: making all our data portable between applications. The good news is that a couple of open standards have already been introduced. OpenId, OAuth and OpenSocial provide the technology to make a big part of our data portable across multiple applications. But an average user should not be involved with any of those technologies. The technologies need to be utilized to build the ’services of the Social Web’. Namely: Identity Providers, Social Graph Providers and Content Aggregators (like Plaxo Pulse, FriendFeed). It might even be possible to make the social graph part of your online identity, leaving the need for a separate Social Graph provider.

The services of the Social Web still leave some problems unsolved. For example, they don’t provide a solution for the management of your online reputation. I would like to take my Ebay reputation with me when I register at Craigslist. Or use my slashdot karma when I go to digg. It’s a tough one since the context often matters for reputations, but most certainly a step we have to take in order to open up the social web.