A revised TwimeMachine
Tuesday, December 6th, 2011A few months ago, I decided it is time to sell TwimeMachine. The code, 2 years old and ducktaped to hell, was breaking. Random errors kept slipping in. Instead of working to fix it, I rather wanted to just sell it and work on something new. Just as I started looking for buyers, the ad revenue and traffic shot up. I was rather busy and couldn’t fix it. Time went on and I had to relearn Django for test. Opting to kill two birds with one stone, I recoded TwimeMachine in Django.
As per usual, when I start a new app/web idea I try do it in a new language, with new architecture, whatever. Just something new I can learn. So, I opted to learn how Heroku works as well Amazon’s S3 for serving static files. It was quite a learning experience, learning how to set up virtual environments for Python and getting Heroku to work together.
So, what’s new? Feature-wise. Nothing new. But I have coded in such a way to easily add new features. I am probably going to add a once-off premium service in the feature that will enable users to download their tweets, read @mentions and read hashtags. When that will, I don’t know. Architecture-wise, it is running on Heroku and serving static files from S3. Will see how it pans out. I still have no clue how to determine when to scale to more dynos. Apparently one dyno can handle between 10-50 requests a second, which amount at 10 requests a second to about 864 000 requests a day. So far I’m well within that range.
Heroku offers 750 free hours, which if I am thinking about this correctly, amounts to 31 days of constant requests (at 1 request per second). That amounts to +- 2.6 million requests per month. My calculations might be wrong. If I am, please correct me. I’ve tried googling for these stats, but haven’t found any. As for S3, it should basically be free.
So currently, I’m running it all for free. Which is awesome! Lastly, I have set up opt-out following of @twimemachine. I get around 300-600 uniques a day. So hopefully I can build some traction from that into my Twitter account. Currently, nothing on TwimeMachine has any traction. There are no network effects or anything being stored (atm), so I’m relying on that to build up some value.
So, go to TwimeMachine and have fun reading your old tweets. And remember, it is not cool to live in your past.
If there are any errors, do let me know.
Cheers!