This post will discuss planning and engineering involved in taking a Website Screenshot of every two-letter Domain. In addition I will share some analytics regarding the traffic Hacker News sent my way.

The original idea

I originally had this idea "Dec 5, 2012 at 10:23 pm" according to my Trello board. I had a number of tasks and hurdles to solve before I was ready to tackle the idea.

Tasks and hurdles

  1. First, I didn't even have a real blog system. I was adding blog divs to a static mako template. I switched over to using Pelican static site generator "January 06, 2013 11:54 AM" according to the blog post.

  2. Next, I needed a way to request screenshot images just-in-time otherwise the bandwidth usage and the LinkPeek website screenshot service would have been at capacity after only a few page views. For that I developed hover-lib which is a small css/javascript library which handles most of the heavy lifting. If you need just-in-time on-hover images, check out hover-lib.

  3. Finally, I needed to write the page. For that I used the following jinja2 template and custom LinkPeek filter (which is just the python function found here - liblinkpeek.py):

    {% set pool = 'abcdefghijklmnopqrstuvwxyz1234567890' %}
    
    {% for drip in pool %}
      <h3>{{ drip }}</h3>
      {% for drop in pool %}
        {% set target = drip+drop+'.com' %}
        <a href="http://{{ target }}" class="hover-lib"
           id="{{target|linkpeek(size='500')}}" target="_blank"
           rel="nofollow">{{target}}</a>
      {% endfor %}
      <br />
      <br />
    {% endfor %}
    

    Above we loop over a pool of letters and numbers, create an h3 header for each section, create a link for every domain, set the link class to hover-lib, and set the link ID to the api call (which hover-lib uses for on-hover images).

Analytics

  • The page reached position #5 on Hacker News
  • Max concurrent and active users was 274
  • Average concurrent and active users was 181
  • Sunday had 7,432 unique visits
  • Monday had 4,038 unique visits
  • We served over 182,025 website screenshot images during the two days
  • No sign ups! (appearently my target market is not Hacker News)

Graphs

CPU usage -

website-screenshot-every-domain-cpu

Network usage -

website-screenshot-every-domain-network

IO usage -

website-screenshot-every-domain-io

Miscellaneous Thoughts

  1. People reminded me that I only covered the .com TLD. Perhaps I will follow up on other TLDs at some point (the code only needs a few changes).
  2. This idea was both driven as a learning experience and as a way to get LinkPeek's name out to the world. I think it was successful.
  3. In the next few days the traffic to the page will converge to zero. This type of hype or linkbait project is like a firework display. It takes lots of planning and engineering to setup and in a moments time it becomes consumed and forgotten.
  4. I need to start coming up with ideas that are not "firework displays" so that the energy I put into the project continues for years.
Browse the blog archives or subscribe to the full-text feed
.


Let us capture, store, and update webpage screenshots for you.

See Plans and Pricing
No Risks, Pay-as-you-go