Twumult is out in the ‘most dirty’ version

You are about to witness the first public appearance of Twumult. We’ve been laying the basis and have just finished the ‘most dirty’ version of the service.

Within this version you need to login with your Twitter credentials. After putting in these credentials you will be able to add other twitter users directly to your dashboard. At the moment it only supports one tab. In the future you will be able to create more tabs and create groups. After creating groups filtering with Twitter Search will be added.

Here are some examples you could build with this tabbed interface:

• All your favorite people on Twitter from Amsterdam, Rotterdam or The Hague
• Your co-workers on Twitter
• People attending to #leweb (add the people you know that attend and add the hashtag to see what others say)

So this is the ‘most dirty’ version of our Twumult which means your looking at a proof of concept. Features coming up are:

• Define groups of user per tab
• Use Twitter Search to follow hashtags as a addition to a group
• Ideas from you Drop them here. Or leave them as a comment.

For now have fun playing around. In the meanwhile we’ll play a song on Guitar Hero!

What we are trying to learn

With our last mashup (Tweetburner) we’ve blogged about the usage of the service. We did this to show what the first days of the mashup did to our server and visitor statistics. A lot of people loved this and, as we hope, learned from it. Last week we’ve been thinking on what we could share with the community this time.

During the start of Tweetburner, a lot of users and people at conferences asked that one question: ‘What’s your businessmodel’. We didn’t have one at that time. We sure could think of multiple models. Most of the times the conversation was continued with the question: ‘but it’s costing you money and time, right?’. But of course. We spend a lot of time building it and we needed to host the service. Which costs us money. So most of the times we ended up explaining why we build the service. Our primary purpose was to build a mashup in a weekend and see how it evolved in the early weeks of it’s life. We noticed that the community was interested in this experience, so we shared this via our blog.

Now we want to share our experiences again. This time we want to find out how to make a mashup cost-covering.

To make some money we need a simple ‘businessmodel’. Google AdSense will be integrated in the interface. These ads can be removed by donating. Donating is starting at $5. A donation will disable the ads forever. In addition to that we will add a donation button to the website where users can donate at any time. We will publish a list with the names of the people that donated on our website. Kindly inspired by the people of Phusion Passenger.

More information on the ‘businessmodel’ will be published soon.

De-noise your Twitter with Twumult

It’s time for the unveilance of our weekend project. Yesterday we started a little contest. ‘Guess what we’re going to build and win a-still-to-be-designed mashup tee’. We got seven guesses, and the eighth one just got in. Later today we will announce the winners!

Meet Twumult

There we go. That’s the name of our mashup we’re going to build this weekend. Twumult will be a way to ‘de-noise your twitter’.

Where the timeline is build for conversations it doesn’t help you much when you want to focus on just a few friends or co-workers. Some applications solve this problem in their own way, Tweetdeck for example. How much I love that application, it came apparent that it didn’t solve the problem completely because it kept the timeline.

Twumult won’t use a timeline. It will be iGoogle meets Twitter. We create a dashboard where you can group your followers and see three (or more) of their latest Tweets. In addition we’ll integrate the Twitter Search API so you can apply filters to your groups and follow people and hashtags on conferences.

That’s it for now. Michiel and I will post a video about the inspiration and development somewhere today. So we can talk more about the concept in person

T-shirt competition closed

In total we have received seven guesses to our mashup. Some are very close to what the idea for the mashup is. The winners of a t-shirt will be announced later today!

What’s next?

We just finished the Amazon images so the server is ready to eat some code. Currently we’re building the landingspage so we can announce more about the mashup. We hope to launch the page around 1 p.m. Then we will move on to build the first rough prototype of the application. Our aim is to have something online around 6 p.m.

In the meantime we keep you informed about our progress via Twitter, this blog and Seesmic movies.

How are we going to run this thing?

Being scalable is important. Especially when you’re aiming at a large (or small!) amount of users for your new web service. Being able to add power when it is nessesary and remove power when usage is low is key for a flexible hosting and development process when building a web app.

Since our app is such an app with an uncertain future in visitors, we decided to use Amazon Web Services from the start. We at least learned this much when building and running Tweetburner

So what is the server setup?

It’s quite simple: we created a server image based on the Alestic Ubuntu images. The current Twumult image runs Ubuntu 8.04 LTS Hardy. Ubuntu is a great distribution for these apps since there are a lot of tools already available and due to the large community and Debian base practically everything you need is available as a package for you to install.

The reason we build our custom app image is so we can scale up whenever we can. It makes it possible to press a button, change some configuration files and you have a second server up and running with all software configured to run the web app. If you would use standard images for every deployment, you would have to re-install them every time you add an instance.

And your database? EC2 instance data is not permanent, right?

Right. If you terminate an EC2 instance all your data will be lost. For running a persistant MySQL database is this not what you want. That’s why Amazon introduced Elastic Block storage. See this as a USB harddrive connected to your virtual server. If the server dies, you can take the harddrive with you. You store your MySQL databases on the harddrive and can take it with you to any instance you want MySQL to run on. Furthermore,  you can expand the size of the Elastic Block Storage on the fly!

What about software?

As said before, we run everything on Ubuntu LTS. Besides that, here’s the list:

• Apache 2 for being the ultimate gatemaster
• Ruby on Rails
• Phusion Passenger for serving Rails
• Capistrano for deployment
• Beanstalk for Subversion hosting (no, not on git yet)

And a lot of other small stuff like logrotate to rotate our production logs.

I hope you liked this post. Do you have feedback, comments or questions? Let us know

Guess what we’re going to build this weekend

As some of you might know, we’re going to build another mashup this weekend.

Currently, the idea is still rough. We spend some time on thinking and made it to a global concept. During the weekend it probably shape up and will be different compared to the idea we have right now. We hardly can’t wait to tomorrow morning. To have a bit fun in advance we came up with this little competition.

Guess what we’re going to build and win a still-to-be-designed startup tee!

We throw in a little help. We will build it within one weekend, and it will be a Twitter mashup. And to give a little more direction, here are two hints:

1. Twitter is about people

2. Stop time traveling

Submit your guess as a comment to this blogpost. We will review them all. Five t-shirts will be available. If there are more than five correct answers we’ll raffle five winners!

Submit your guess before 10 a.m. tomorrow. After 10 a.m. we announce the mashup!