People are looking for free hosting

Looking at search queries leading to our website i find “free php hosting” to be the top five most frequently referrer. Also, Google AdWords constantly suggests to bid on keywords like “free hosting”. And, we see that people are especially attracted by the free fact:

Fortrabbit – PHP Hosting Platform | PHP 5.4, Git, SSH, Memcache, Mysql myshar.es/WC4MAF Free #hosting for #php Awesome!

— David Higgins (@_higg) 24 décembre 2012

But hey, we don’t do shady business here! Our platform is a really sophisticated high quality hosting service.

Free web hosting origins

Time warp to the 90s. Until then only big companies could afford to present themselves in the internet. Some universities offered webspace for their students. And the first Hosting services for end consumers came up: Services like GeoCities, TriPod and AngelFire offered free hosting while displaying ad banners. My theory is that this somehow became Facebook and Tumblr later on, but that’s a different story. A bit later classical hosting was established. It offered inexpensive shared hosting plans. Fake-free: Providers often used “zero price” tags to convey customers that their plans are for free.

Freemium

Nowadays users are aware of subscription traps. Crippleware to the rescue. Freemium is a low-touch marketing trick: A service provider gives away a core part of the product for free while trying to push the user into a paid plan.

For Software as a Service the freemium level is mostly fictional: the operating costs are nearly zero. The service is worth … well … what it is worth to you. It save you time, it saves you money, it let’s you be more productive, you can do things in a different way, or you can even do things that you haven’t been able to do before. Great. Awesome. Basically SaaS providers just have to make sure that the conversation rate from freemium to premium is high enough.

For our Platform as a Service the same benefit values apply. But we also rent out computing resources and those always cost some real money. That’s why it is so hard to offer a real good freemium product here. We want to be as open as possible, let people experience the benefits of our service, but the paid plans should not pay (too much) for these free tiers. We calculated very carefully how much we can spend on this. The result is that Apps running on a free plan are freezing after 48 hours of inactivity. Now, we see that people are annoyed by that:

Fortrabbit I am trying to like u, but if u keep on freezing my app I regret to say I won’t like you. servers dont run on trees @fortrabbit

— Andrew Smith (@silentworks) 10 octobre 2012

And we can understand it. Our aim is to provide a painless PHP cloud development environment for free. In an ideal world we would like our clients to convert to a paid plan just because they love the service sooooooo much, just like we all purchased a Sublime Text licence, because it is such a nice editor.

We see our PaaS colleagues struggle with this as well. Appfog recently limited their free (forever) offering. DotCloud has even canceled their freemium plan completely.

Quality PHP hosting can’t be free. We are evaluating very carefully if and how we can make our free entry level better.

@robjmills @fortrabbit Well in fairness, I wouldn’t depend on any free service for a production app. The price is reasonable too.

— Niall O’Brien (@niall_obrien) 5 avril 2013

We are a tech company with a twist for philosophy. These are two of our company main mantras:

Encourage best practice

We really like the way our great PaaS forefather Heroku enforces their users to use best practices (like Git deployment). However, we are more softcore. We like to encourage our users to use best practices. In other words: We support your habits. On fortrabbit you deploy code the way you deem best. We believe that standard compatibility is utterly important. Of course we offer modern tools and techniques (like Composer) for the most sophisticated workflows. And we like you to give them a try with us, if you haven’t used them before, in the hope you will find even better ways to reduce your workload. Our support pages and our interface help you to get started quickly.

Security first

Whenever we have to make a design decision between security and something else (usability or performance for example), the former mostly wins – as long as it does not prohibit any level of usage, that is. We strongly believe that we, as your trusted cloud hosting vendor, have to take security VERY seriously.

For example: we decided to use a very strong encryption for all the service (SSH/SFTP, MySQL, ..) passwords via one-way hashing. As a result not even we know or can access your passwords. However, this comes with two usablity downsides:

1. You cannot read out your service password in our dashboard, if you lost it, you can only create a new one.
2. Most other PaaS providers give you the MySQL credentials via environment variables. Granted, very handy. But it requires to save the password somewhere in plain text, so we can’t do it.

Writable Storage

Most other PaaS don’t offer writable storage and thereby have no need to make it accessible by any kind. When you deploy your App, it get’s copied to the different nodes it is served from. In terms of usability and compatibility we think this is well below ideal. Our users should be able to use their favorite workflows, frameworks and CMS, without jumping through hoops. There should be no hacks needed to run a standard software such as WordPress. That’s why we provided a shared storage for each App, to which you can deploy your App with Git – but also via SSH/SFTP.

However, there is also another usability need in the context of writable storage: your App should be able to write to the file-system as well, Think: user uploads, CMS web based upgrades, online editing in themes and so on. So far so good for compatibility and usability.

The Catch

Of course there is a catch: Allowing your whole App to write on the storage, as we did and most of the other PaaS do, leads your App open to attacks from a certain vector: your very App itself. If there is a security vulnerability, not necessarily in your code, but in the framework or CMS part, which allows uploading .php files, an attacker can use this to inject mal-code into your App and then .. can do whatever he wants.

On the lower level security side, we already took great care to truly and fully separated each App from another to limit any effect any compromised App has towards all the others.

The Solution

Recently our user Francisco Azevedo pointed out that we could (and should) grant more control of the level of security our users have over their Apps. He wondered why there is no way to limit write access for Apps and suggested that Apps might even have no write access at all per default.

We thought about this and created our new Write Protection feature. We think that this implementation brings best of both worlds together: By default write protection is off. This is important to assure compatibility and also most Apps start out in development. It is essential that new fortrabbit users get their stuff up and running easily, without obstacles. We think that Apps in development (mostly in the freemium plan) don’t need write protection necessarily. Of course you can turn on (or off) write protection at any time in the web interface. When you scale your App to a production level plan and write protection is off, you will be presented with a very visible warning message that is a good idea to finally turn it on now.

Our Learnings

This issue touched us deeply. You can’t imagine how much effort we put in designing the platforms infrastructure architecture. We think that it is the best thing we have ever done. Yet there was still this important detail that we haven’t thought through enough. Our learning is that we should always be on the guard and question all of our moves as carefully as we could.

Thanks again to Francisco for this very valuable feedback. It is a great honor for us to have such smart users who also really care so much.

Symphony is a content management system powered by XML an XSLT. It’s quite mature, well tested (2.0 was released in 2008, current stable is 2.3.1) and focused on simplicity in design and complete control for the developers. Symphony is completely open source, but commercial support is offered. There is a huge set of extensions and macros in form of XSLT Utilities.

Currently, Symphony is migrating towards PHP 5.4, so there are still some quirks you have to deal with.

The majority of you explored our platform during the last 6 weeks. We belief 1,000 is just a fraction – there are zillions of other developers who never heard about us.

For us, 3 guys with a simple idea of better platform for PHP developers, this tiny number and your feedback shows that there was something missing in the current PaaS-World. It shows we are on the right way, with the right tools and the right attitude to make make it big.

A big THANK YOU goes out to Bruno Skvorc and Gabriel Manricks. They mentioned us in their articles at phpmaster.com and net.tutsplus.com

Work in the cloud

So, what’s your benefit exactly? If you want, you can give up your localhost now for good. With PHP on the shell, you can run your favorite framework CLIs directly in your App’s SSH account. And there are many:

• artisan from Laravel
• Symfony brings it’s console
• Every framework using Doctrine comes with doctrine-orm
• CakePHP provides the cake command line tool
• Yii Framework comes with yiic
• FuelPHP has oil (what else?)
• Lithium gives you li3

Of course, this is only the beginning. You can now use any Phar tool you like. For example Composer, which is already part of our deployment, can now be used in sub directories. Or have you heard of Phing? Great tool!

But that’s not all: Git on the console helps you with your legacy dependencies (that is: before composer) ⇒ use sub modules or private repos.

Naturally, you still can use your localhost – but don’t have to, if it’s inconvenient.

Limits

Yeah, there are still some. Your App’s SSH account is in a shared environment, and we have to keep up the quality. So you have limited memory (~300MB) and running workers is still not permitted. The alternative would have been to increase base-prices by a lot to assure we can pre-allocate sufficient resources for everybody. But hang in there, there will be dedicated SSH servers in our Enterprise upgrade. With those, you also can schedule cron-jobs and can choose how much memory you want.

• See the roadmap feature
  

More on our wide range of optional PHP settings here and here.

We are busy with our new PHP developer cloud hosting. Currently we are preparing to shut down our old MISH services on the 30^st of April 2013 in favor to the new platform.

A part of MISH is WebRechnung and we finally decided to take it down without replacement. WebRechnung was a free service to create and send invoices and estimates online – similar to FreshBooks, but especially designed to fit to the German tax laws. We have failed to make it a real business. This article is about our lessons learned.

See WebRechnung in action in the video above.

History

The Need

By the beginning of 2010 we needed to generate invoices for our old hosting solution MISH. We looked around and as we found nothing at first glance that looked cool enough for us we decided to build something on our own right away.

The three chief virtues of a programmer are: Laziness, Impatience and Hubris. Larry Wall

It turned out that it became a really useful tool. Our friends and ourselves began using it immediately for any kind of invoicing. So we launched WebRechnung (WebInvoice) as a stand-alone product in September 2010.

The business strategy

It feels a bit strange when i think about it now: WebRechnung could be used stand-alone, but was implemented and tightly intervened with our original MISH hosting solution. The idea was that people who need online invoicing could also use our hosting and vice versa: You only need one login to manage your hosting account and your invoicing – isn’t that cool? Well, maybe not. Most of our hosting clients never logged in to WebRechnung or the other way around  – no real synergy effects.

After we fixed some teething troubles in the first months, WebRechnung soon became a mature product with all essential features. It was just cool like that. We have had some competitors (EasyBill, Salesking, Billomat, FastBill, Collemx, LexLive …) but our product and it’s price where really good, I think. However the sales never really took off. So we neglected it a bit.

Set it free

By the end of 2011 we only had about 50 paying customers with an overall monthly revenue of something like 300 €. We decided that we should focus on our core product hosting and put no more energy in this invoicing tool. We didn’t wanted to feel too responsible for our few paying customers. So we removed all costs (aside from the one we had ourselves: sending invoices via snail mail). So we had a professional developed billing software for free – no handicapped freemium crap.

My secret hope was that maybe now it will have a positive effect on our hosting or the invoice user base could increase to a critical mass so that we could think about a business model again. More contacts should help to manifest the fortrabbit brand, improve our Google ranking and so on. With enough users we might have even been able to bring back some new “premium features”.

Still not enough success

It turned out that even this bold move didn’t made any real impact. About two years after the launch and one year after removing all costs from it we still only had about 200 users really using it.

Sometimes it’s not enough to have a good service for an unbeatable price. You must be visible in a competitive market. We didn’t had any marketing, nor social media activities, nor public relations, nor customer relation ship management going on. I think that this was our biggest mistake.

Seeking possibilities to continue

Our main problem with WebRechnung is that it doesn’t fit our standards any more – software, as everything, degrades over time. The interwebs is moving fastly forward, everything is evolving. Adapt or die. We don’t want to see this old piece of software (base core is 3 years old now) out there into the wild much longer.

What about Open Sourcing it?

We would love to publish it on GitHub. Unfortunately this software is not really a modern ninja style stand alone application. It is tightly integrated into our monolithic MISH system – mostly written in Perl and impossible to understand for an outsider anyways. And it’s really old: dependencies are outdated, large parts of the code are in need of major re-factoring. We would rather rewrite everything before touching this code again.

What about crowdfunding it?

I thought about a kickstarter campaign for a while. But we are into this new PHP developer hosting platform now and that is actually much more interesting for us. We need to focus, we don’t have the manpower to develop and maintain such a huge side project.

Dear existing clients, i am personally very sorry to bring you such bad news. It was a really hard decision for us, but i hope you can understand us after reading this. The good news is, that there really similar services out there. I have named some on our product site: webrechnung.info. The beginning of the year is a good time to change your invoice tool.

We had joy, we had fun, we had seasons in the sun.

Some more comments on HN.

Current Traction

During BETA and the first weeks after launch PHP developers from all over the world were paying attention. Early adoptors, some of the Top of the PHP Pops, web agencies and even competitors tweeted, signed up, played around and gave valuable feedback. But what’s most exciting: People love the product we’ve build, they pay money for the service and use it in production.

We know that PaaS is still under the radar of many PHP developers. But in 2013 developers, organizations and enterprises will start to rethink their current hosting situation and take advantage of managed cloud environments.

The Road Ahead

The awesome feedback of our customers is the best inspiration for us to build and develop a service that solves real world problems. They ask the right questions and give us hints to make the platform better. So, please don’t stop to ask and tell us your ideas. Here is a short list of out upcoming features:

• Optional PHP Extensions
• PHP Runtime in SSH
• Enterprise Products
• and more if you like

Company Building

Our line-up of three self-motivated guys worked very well to ramp up a very solid core product and we know now it takes time to form a company. Building a team that allows us to make bigger steps and to keep the support on a high level, without losing the fun will be the biggest challenge. It’s a shift from a hacker to an entrepreneur.

Bootstrapping the first development cycle was a great way to start, but we know that good staff expects a solid salary, so we have to bridge the gap until being profitable. Company funding is a kind of unknown territory for us (we are engineers/designers). Fortunately there are different options we can explore and there are some advisors around guiding us.

As you can see, there are many things that will keep us busy in 2013.
Let’s go for it!

Web Service

Monolithic frameworks are out, component based collections like the new Laraval 4 and Symfony are a better approach. I believe that the same light weight set up could apply to our hosting and web services as well. The old price dumping mass hosting services sucks and must be replaced. There are great new specialized services for your needs:

• Simple CMS: Squarespace, Virb …
• Portfolios: Cargo, Carbonmade
• Blogging: WordPress.com, Tumblr, Blogger, Scriptogr.am
• eCommerce: Shopify, Big Cartel
• Static Pages: GitHub, Amazon S3
• Personal Splash Pages: Zerply, About.me, Falvors.me
• Professional Developing: Heroku, Appfog, Dotcloud, Pagodabox, Cloud Control, Fortrabbit and even more

But to have your business up in the interwebs you will most likely need a memorizable addreess, a TLD. And you might also need to send and receive mails from this address. All the services above offer you a way to connect your domain, but that’s it no service will host your domain nor your mails. So where to go then?

In decoupled thinking domain and e-mail are two different things. So we are looking for two services: One to order and transfer domains and to manage DNS settings. One to set up and configure new mail addresses and to receive, SPAM filter, store and send mails from.

Domain Service

It’s really hard to find a service that does only domain hosting out there. Domains are usually not a product of it’s own. They are often part of old school hosting plans, you know the ones that start with check domain name availability now. However some hosters offer dedicated domain (and mail) packages. Luckily there are some new specialized services: iWantMyName and DNSimple just doing domain management. Both services match perfectly to decoupled thinking and the above services. In Germany United Domains, Schlundtech are also single purpose domains registration services, a bit more old school. Domain Factory and Host Europe offer packages with Domain AND Mail.

E-Mail Service

That’s more complicated. E-Mail hosting must be very reliable, SPAM has to be handled. The big elephant in this room is Google with Apps for Business – that’s just like Gmail, but for your own domain. Google is international and will probably never go down. It also comes with extra features: Calendar, Docs and Drive. There was even a freemium entry level, but the have just recently skipped that. Let me guess: They already destroyed the market and rule it now. Left competitors are: Zoho Mail, Rackspace Email, Fastmail (all in the US).

The usual extra business features for business e-mail are fine, but i am not quite sure if i will need them right now.

Conclusion

I would like to see more specialized professional services for domain management and especially for email hosting – E-Mail as a Service and Domain as a Service offers.

Our old hosting system MISH included Mail and Domain handling. These things kept us busy, so i am actually pretty happy that we don’t have to care about all this again on the new platform.

As the old system is closing in April 2013, i have to tell our existing clients how to go on now. Some of them might proceed with classical hosting, that’s ok, see my previous post for this. Others will hopefully migrate to the new platform. I would like to give them directions like this: »We are your web hosting partner, register your domains there, host your mails over there. All services are separated but fit well together.«

1) App Collaboration

Grant your team mates access to your App – multi-tenancy workflows for the web control panel. Set the level of permissions based on roles:

• Analyst: read only
• Junior Developer: change settings
• Lead Developer: change settings, scale App, purchase Add-Ons
• Project Manager: change settings, scale App, purchase Add-Ons, handle collaborators
• Owner: change settings, scale App, purchase Add-Ons, handle collaborators, delete the App

Sorry, this new feature is only available for paid plans. It was simply easier to implement like this.

Who Benefits?

We really like this feature, because it’s not yet another tech feature, it solves a real world problem between people. It helps anyone who has to manage multiple Apps with different co-workers and maybe even different owners. It’s a good solution for dev shops and web agencies.

2) App Transfer

Change the payee of an App by inviting a new owner. The new owner will be guided thru the boarding process. After the transfer you can still work as a collaborator on the App – of course, when both parties agree.

Who Benefits?

Again this helps to manage your real world hosting needs. Imagine this workflow for you as a freelancer:

1. develop a project for a client (maybe on a free plan)
2. send over testing URL
3. get feedback
4. implement changes, send over for review again (loop steps 3 & 4 a few times)
5. get approval from the client
6. invite the client to take over App ownership
7. the client becomes fortrabbit client
8. you are still the project manager and can handle the App for the client
9. you launch the app for the client (scale and route domain)

3) App History

Each app has an event log now in the web control panel. Here you can see exactly who changed what and when.

(4) Bonus feature: Gravatar support

Ok. That’s really a minor change and not really worth mentioning: Now you will find icons on your profile and anywhere accounts need to identified  in control panel. These profile icons come from the free Gravatar service. So in order you want to see your beautiful face up there you need to have your e-mail registered on Gravatar. We have noticed that a lot of our users are already using this anyways.