Edit: I have added my conclusions at the end of this article, read the results of the experiment.
This article is about how I am testing a business idea of a service to notify you when your website or service is down. It also presents the free services I use as a Javascript freelance developer.
Are my URLs always UP? Why not?
As many freelancers, I host services for my clients or for me as demos. As many developers I host them for free and with great performance on heroku or CloudControl for nodejs or other static HTML hosts like the excellent github pages, Netlify, Forge, Divshot and Bitballoon.
There are many great free hosts for static sites and Javascript apps.
In my apps I use other services which I use for free until my client's app reaches the limit of the freemium model. For example I use Mailchimp and its very simple API to store email adresses, Openredis (redis hosting with a free plan) for session data and more,
So many things can go wrong from the domain name to the hosting to the app to the third party services and to me renaming a folder somewhere. And I learned that sometimes my URLs are down no matter what choices I make.
Sometimes my URLs are down no matter what choices I make.
The tools to monitor a service
The tools I have used for now to check my URLs are newrelic, monitis, pingdom. They all have free plans, with limited number of *everything*. They also offer a dashboard with great analytic data, graphs about the response time, the lag, the total load time, in function of the country, with tons of settings... Most of them also offer libraries or SDKs you can add to your app so that they tell you how your app is performing and how your server is doing for profiling, debugging, ... Trace route is a cool feature too when you have a demanding client with unlimited funds - I used it while working at Ebuzzing, where a downtime in one country for a short time can result in a fair amount of money.
There are many services to monitor web servers, but for simple HTML pages it is too complex and expensive.
With these services, when you exceed one of the numerous limits, the service goes quickly from $0 to $1000+ per year, almost instantly! See the pricing of monitis, pingdom and newrelic.
Now what? Monitoshi?!
I need to monitor my URLs: my about me page, Silex Labs numerous websites, and more importantly my clients web apps. For analytics and big data I use Piwik, which is enough for me. So by "monitor" I mean poll a URL and send me an email if something goes wrong. That should not cost me money should it?
So I spent a week-end building something for me, in node.js, hosted for free, simple enough to set up and with a small code base to be able maintain the code over time.
I promised some people that I would consider trying to make money with my little projects (big up @julien), not because I believe I can be rich with a project named "Monitoshi" but because a service can only be good and sustainable if it is profitable (talking about a service here, not a project!). I came to think how I could offer hosting for this project - for which, as a free software fan, I open sourced the code of course.
What a great idea, and a simple project, let's do it ! Oh wait. But why ?
Before starting to work on a website, an admin interface, setting up a scalable environment and try to crowd fund it, I wanted to be shure that this is worth a shot. So I followed the "3 Steps to Validate Your Business Idea For FREE…almost" method (thank you @ariel)
- landing page: here is what I did in an hour, with Silex (free), starting with a template (free), hosted on github pages (free)
- ad campaign: I bought $200 of linkedin ads so that they display a link to the landing page - note that I have tried to setup a duck duck go campaign, a bing campaign and an adwords campaign, and they turn out to be way too complicated for me
- analyse results: I plugged the landing page with a Piwik instance, and I am waiting for results
Tell me what you think of this idea ? Will the results of my experiment be good?
Edit: the results of the experiment
I had very few visits coming from my 200€ of linkedin ads. Around 1 visit per day in the 2 months of activity, with a very high bounce rate (85%). I had more people after I gave up the ads, around 2 or 3 per day coming from the alternative.to page, with a better bounce rate (77%). It is still not a lot.
My business idea failed the test, so here it is, available as a free and open source project for my fellow developers.
Either nobody cares about free and easy monitoring of HTML pages, or I was not good at this exercise. I have developed the product anyway, as a free and open source project for my fellow developers.
On the official monitoshi landing page, you will find a simple form to add a monitor, and a link to the source code. Anyone can host a monitoring service like mine, which is running on heroku for free.