Archive of May 2008

How to Price Web Apps?

I've been thinking about this for the last couple of days as I sit here I work through my C Programming Book. In a couple of days I will be starting work on my yet to be named Web Hosting Control Panel project. The basic idea with this is to build a better, scaleble control panel, license the code under the GNU GPL and sell support services. Compared to all the other control panels on the market this is a completely new idea, in concept and in application. I am unfortunately having problems planning out the support part.

The coding on the other hand is quite clear in my head and has been for quite a while.

The problem lies in the market it is going to be slot into. There are no such projects like it. On top of this most of the web hosts that I have come across on the forum "Web Hosting Talk" are quite frankly scared of open source. This among other things has taken me to decide to sell *optional* support to people who need it. In my minds eye, this is the upshot of this model and it gives me the most flexibility.

My problem lies in what to charge. I initially came up with three package ideas. These were: Single, Server and Site. These would provide support on however many installs there would be per machine. In context a "Single" would have a single install, "Server" up to as many as 30 (as a VPS Node) and "Site" which would provide support for all the servers in a single building. These would therefore be priced accordingly.

The trouble I have found with this model however is that I cannot easily come to an appropriate price. Depending on the install, the "Site" package for example could give me almost unlimited support requests to the point that I could not cope with them on my own and then would have to rush in help from elsewhere. Not an easy task for a startup.

This gives two problems. Pricing too low and Pricing too high. If I price any of the packages too low then people will not take it as a serious control panel project. However if I price it too high I will not get any sales - sales which are important for the project to stay alive. With this in mind I have to consider pricing it at a level where I can expect support to lie. This is where the root of having the three packages comes into play, but circles back to the pricing problem again.

As I am considering all this before I have produced something I cannot, realistically price it as I do not know how much time went into it as a project. And as is typically mentioned in such articles "time is money". Going by a reasonable wage at £50/day and then multiplying that by the amount of time I can allocate that brings me to some £4,900. This also assumes an 18 hour work day - from getting up at 9am to going to bed at 2am the next morning. That's also less than the UK's minimum wage.

All that of course is just labour costs, and that is just for the initial development period. So that is a ball park figure which in effect does not represent anything other than my own labour cost.

Since starting this article I have thought about both monthly and one-off pricing. Given that support is an ongoing service, like a water supply is it is not appropriate and would have to rely on a constant amount of big sales to make it profitable.

The general given monthly prices for single installs of the main panels are show below:

• cPanel: £10/m
• DirectAdmin: £10/m
• Plesk: £20/m

Note: I must say that these are from memory, and are most probably not correct.

However, looking at these it shows that for licenses for single installs are quite low, but still add quite an overhead for indivual installs which would be, for example on a Virtual Server. Given that my product would be free on its own, it would be optional to pay for support. With of course the general hope that people do pay for support.

To break down considerations for each, I've split this into sections.

Single:

• Must be affordable for a single server.
• Must represent the amount of time which can be expected to be spent on each customer.

General Price Range: £10-£30.

Aim: People running single servers with only one install.

Server:

• Must represent the amount of single packages availible per server, but also be more affordable.
• Should not undercut the amount of time that will be spent working with the customer.

General Price Range: £100-300.

Aim: Companies running VPS nodes, and /or large amounts of resellers per server.

Site:

• Must be priced at a level which is attractive to much larger customers but in balance so it is a steal compared to the first package.
• Must realise how much work will be involved.

General Price Range: £1000-£2000.

Aim: Datacentre owners, Companies/People renting cages, Corporates.

In Conclusion

I hoped that I could get to this stage earlier as now this is a very large post however it does show the consideration that has to be put into how projects are priced. You also have to remember that these are priced per month, with differences between each market - this is what explains the high prices for the whole site.

Putting together the Development Environment Required to Programme for Pocket PC's

I am just in the planning stage of putting together a utility for my Pocket PC. Seemingly however I could not find any documented article which explained what was required to put together the basics that are required. These are; the IDE (Integrated Development Environment, contains the parts required for the components, GUI tools and Compiler), the SDK and any updates. This is mostly based around Pocket PC/Windows Mobile 2003, not the newer devices.

Note: This article focuses on setting up the environment for C++ Development, not any of the newer .NET tools.

Applications Required:

You will need three apps, which should be installed in the listed order:

• eMbedded Visual C++ 4.0
• eMbedded Visual C++ Service Pack 4
• Pocket PC 2003 SDK

Note, the first links are so you can download them from Microsoft directly, you can download them from me if you are unable to access these (as a zip file). Also note that the main app requires a serial key. (Which is seemingly odd for something downloaded from the Microsoft site).

Once these are setup it is quite easy to create a simple HelloWorld app; for which it will provide you with an example.

The Importance of Documentation in Open Source Software and What to Use

Typically overlooked, documentation is by far the most important part of open source projects. Seemingly the more hardcore programmers negate commenting code and explaining how the code works. Whether it be a Wiki, a pile of HTML pages joined together or a pretty PDF anything is better than nothing.

In this article I am going to compare the ups and downs of each method, hopefully come to a conclusion of which is best and outline my plans for documentation in Four Cores, my primary project.

Wiki's

Wiki's are great, they allow users to interact and overtime develop content. However they are open to abuse and empty Wiki's look like unfinished books. A Wiki is at its basic element a site where people can edit the content given to them. This allows errors to be corrected, other options explored and much more easily kept up to date. Out of the box however they are very much uncustomised and usually look bland and uninviting.

Being the way they are however they are easy to them and customise. The most popular, and what drives Wikipedia is MediaWiki and upon installation it looks exactly like Wikipedia, sans logo's.

The main problem with Wiki's however is drawn from its user-based design. This means that information can be outright wrong, or vandalism can take place. Vandalism of course is mostly common on high profile sites, notably Wikipedia which is a constant target for problems.

Once this is weighed up however, Wiki's provide an all in one location to provide an interactive documentation source which may not only contain text, but can contain other forms of media, such as videos.

HTML

HTML based documentation; a series of files simply linked together is one of the most common types in the open source market. Typically written by project leaders it provides a version dependent source of information in a fast and streamlined manner.

However, this causes problems. As mentioned in the above sentence, HTML based documentation is typically version dependent, this means that once a new version is released the old version needs to be updated, by the original source documentor. This stops the push of development to instead document new features which the users cannot ascertain by themselves.

Although a good multi-platform medium the rise in, and consistent use of multimedia across the web has pushed this into the dark ages, with users expecting to be able to find a much better looking version to sit back and read. Regardless of the content.

PDF's

Adobe's PDF format is an excellent way to publish documentation online with print in mind. By taking the original HTML based information and providing this in a single PDF document it allows for versioning which may not have been seen before in the respect of HTML methods. At each release, this can be updated and provided in PDF form for each version.

The element of print; as in, all the elements stay in the same place whichever machine the document is viewed on, allows for the document to be designed rather than a hastily push together of files.

With their typical use in many forms, the drawback of its format is something that, in effect no longer needs to be worried about.

With Four Cores in Mind

This post has mostly been thought out in preparation of Four Cores, my currently, standby commercial project in mind. My current plans with it are to produce an Open Source, but commercially supported hosting control panel in which providing manuals for its installation and operation will be paramount. Even with its current remit of using the GPL to license the product itself and to produce an income, sell support services for it providing a free manual is important for its status as a worthwhile product.

If manuals were only provided to paying members, the serious uptake would be minimal, something I obviously want to avoid.

With all this in mind I have taken the decision to run both a Wiki and a PDF manual set. Both will be available to paying and non-paying members with the only difference that paying members will be provided with direct support, whereas non-paying will not.

A Case of being Out of Stock, then £20 more.

I'm quite annoyed at ebuyer.

I had planned on getting my case on Monday evening, but I didn't as I hadn't yet won my CPU (eBay auction). Fortunately that night I eventually did win my CPU. It's a P4 2.8 with a 1MB Cache and 800mhz FSB, I'm rather chuffed with my win.

Unfortunately though, when I went to check the next morning, it was out of stock. Irritated I left it until today, thinking that hopefully it would be back in stock. It is.

Its £20 more though. This makes it less economical to buy a kit (of which it is an Antec budget case and an energy efficient "EarthWatts" PSU). This really is quite a set back, mostly because I had expected to be receiving the Case and PSU by the end of this week ready to load everything in and test it, hopefully with it all working out of the box and me not having to flash BIOSes, adjust this and that, etc, etc.

It seems that now my budget is going to have to increase past the £120 mark up to £150 so I can complete my project. This annoys me - it was initially started as a side project to building a Core 2 machine, which although I cannot currently afford, I can afford to do this. It seems though that this is going to push the main project back as I am going to have to either decrease the values of the components, or spend more money. Neither is really the most wonderful option although the latter is something that I would rather entertain than getting less high-spec components. There is of course the option of getting my other server running, but I am yet to find an appropriate location for it. (Due to noise).

I am now going to have to search and search for this case at a reasonable price, or go for the separates option which originally cost £10 more.

Three Good Films

Over the last two nights I have watched three really good films. They have been: There Will Be Blood, about Oil. The Prestige, about two magician and finally The Counterfeiters, about the Nazi operation to make the Allies economy collapse.

I'm not trying to review them here, so what I will say is that all three have been very good, and I'd recommend them.