New startup

[suklee]

A few of us are getting together and planning a new website, which we have big goals for. My experience is in PHP/MySQL and a few of the other guys are Microsoft/MsSQL people and sent along this http://www.microsoft.com/web/websitespark/

I was under the impression that using Microsoft products to build, host, and maintain a website would "cost" more, but does everything just equal out in the long run? (versus using open source technologies)

How does one go about deciding which platform to build our site on? And moreover, does it matter?

Thanks in advance for your input.

 

[degibson]

Originally posted by: Kai920
I was under the impression that using Microsoft products to build, host, and maintain a website would "cost" more, but does everything just equal out in the long run? (versus using open source technologies)

I'm sure M$ has spent a lot of money to sell the idea that TCO of an M$ toolchain is lower or equivalent. Honestly, it all depends on the people -- what toolchains the people prefer, have been trained on, know and love. Training people is usually expensive. Perhaps not so much in your case.

How does one go about deciding which platform to build our site on?

First, figure out what your scalability requirements are. Benchmark your options, and lump them into two categories: what will work, and what won't work.

If multiple choices exist that will work functionally, pick the one with lowest TCO. Unless you're n00bs on open source software, that'll be the free stuff (assuming the free stuff scales to where you need it to go out of the box).

And moreover, does it matter?

Yes, it does. When it comes time to innovate, its nice to have made wise decisions about your infrastructure.

 

[suklee]

Thanks for your input.

Myself, I lean towards open source but I'm keeping an open mind. My other partners though seem to be set on MS - they have the conception that free, open source software can't be used to power an enterprise-level web platform. That's my gut feeling any way.

How does one 'benchmark' the different options anyway? (And what are some of the different options?)

 

[Markbnj]

I did a startup with three friends in 1995, which we ran pretty successfully until 2004. We are all primarily MS developers so that is what we based our platform on. There are potentially multiple aspects of this question that you have to consider.

There's the internal view: how much will it cost to get started? What are our people familiar with? What will be the costs in the long run? How fast can we move in the short term? These questions apply not just to platform choices, btw, but also to development practices and SDLC.

There's also the external view, which varies dependening on the business you're in. What your prospective customers are used to, what they expect and respect, what is already approved and what isn't, should have a large bearing on your decision. We entered into a primarily Unix server-oriented market with a solution built on MS, and it made things much harder in the early going, as we were opposed by in-house IT every step of the way.

 

[degibson]

Originally posted by: Kai920
Myself, I lean towards open source but I'm keeping an open mind. My other partners though seem to be set on MS - they have the conception that free, open source software can't be used to power an enterprise-level web platform. That's my gut feeling any way.

Ask your partners if they think M$'s major online competitors use M$'s products in their datacenters. In other words, there are plenty of obvious enterprise examples of non-M$ web platforms.

How does one 'benchmark' the different options anyway? (And what are some of the different options?)

Well, its hard to be specific without knowing what you're doing -- and you probably shouldn't post what you're doing on a forum. But if this was 2002, and you were about to invent Facebook, you would want a web infrastructure that could scale to thousands of simultaneous frontend users, hitting java middleware, touching a huge backend storage system. The middleware would be your own innovation, probably, but you'd want to look around for the right JVM. The frontend would be either Apache or IIS (probably Apache)... and as for the backend, who knows? One might be tempted to go for a distributed database, but you don't really need ACID for the workload... so you could homebrew, or look for something weaker and with better scalability.

The point is, you have to think very hard about your own requirements to figure out what you need -- then you need to figure out how to benchmark the things you need.

EDIT: Like many startups, I assume your goal is to be purchased by a larger company... so your immediate goal is probably proof of concept rather than long-term viability. The scalablity part still matters -- but expect your initial implementation to be thrown away if you're acquired.

 

[brianmanahan]

Sorry degibson, saying "M$" is so 1990's...

 

[GodlessAstronomer]

Originally posted by: brianmanahan
Sorry degibson, saying "M$" is so 1990's...

I stopped reading both of his posts the moment I saw that. Grow up.

 

[degibson]

Originally posted by: GodlessAstronomer

Originally posted by: brianmanahan
Sorry degibson, saying "M$" is so 1990's...

I stopped reading both of his posts the moment I saw that. Grow up.

Yes, please ignore the OP's question and comments about benchmarking, cost-effectiveness, etc., and focus on my choice of abbreviation. The internet is, after all, entertainment first and foremost.

 

[VinylxScratches]

You could use C#/ASP.NET + MySQL to reduce costs....

 

[aceO07]

Website? Simple web app with cool/useful concept or does it do heavy processing?

If it doesn't require anything specialized to do the heavy processing, I'd say go with whatever you guys know best and get it out fast. Even if it does require something special, you can always just make a call out to a service to do it. Just develop it so it can handle changes without lots of work.

Most people will fail on their first attempt and some will get back up and try again.

[I have NOT done any startups.]

 

[brandonb]

In my experience: With open source software, you get what you pay for.

There are a few free databases, but trying to struggle with the issues caused by them will cause alot of frustration in the end.

If you uue Cache db (which is cheaper but not free) you try to do an inner join to the same table twice (with aliases), you will cause a deadlock with the sql statement (the one join will lock the table so the other join won't be able to access it)...

Postgres has some decent tools, but you run into the same type of issues. Some things which can be done very easily with MSSQL are just a chore to figure out (if at all possible) with Postgres. Usuaully I have to make an external program just to work around issues I can't get done in their SQL db which would have taken me 30 seconds to do in MSSQL.

MySQL experience was limited (a few years ago) but during my time experiencing it, their database management tools would crash all the time to the point I just wanted to pull my hair out and decided to just go back to MSSQL.

I'm not saying that open source doesn't have its place, but if you want to seriously consider a business, you should stick with things that have a price tag.

Another thing to keep in mind, MS stuff has big teams working on it, its used alot, and they have very solid testing (unit) teams. The stuff usually works and if there are hiccups, chances are Google will get you through it because its been figured out by 12 people before you (who wish to post about it). Open source software has that but to a lesser degree. Alot of the time, you post on their sourceforge.net website and hope they respond in a timely manner and usually they don't. While you wait for the response, you twiddle your thumbs or just work around the issue by other means.

YMMV. I usually try to stick with MS products these days after having so many problems with open source products. Some people at my job insist that MS is a big evil corporation and will do things in other tools but they are much less productive (in my experience.) I can kick out code that works while they are working on their "small apps" (What you would think is a few month app) years later trying to get it to work properly and report the right results.

I am their supervisor so I know, and now my company has a policy to only develop in MS products. And yet they are still trying to convince me to use Ruby for web server apps. I just respond "You code in notepad.exe and you can't run a debugger through your code, how much time do you put into your app with print statements trying to narrow down problems? If you have a big app, you will need a debugger. Sure Ruby is fine if it's just a small one page program, but for anything more, you need big boy tools."

 

[sourceninja]

Are you seriously trying to imply there are no good IDE's for ruby? That there is no easy ways to debug or test code written with ruby?

My job is just the opposite. Everything we use is open source, everything works. Our timelines are always met, we are fast to market, and our costs are very low. After I pushed for and won my opinion we have cut operating costs by almost $50,000 year, and we are just getting started. We use MySQL almost exclusively with only one proprietary app that requires us to have oracle (and our oracle license only allows that one app to save cost). We have replaced everything with linux, and do all our web development with a combo of java and php (actually exploring python and ruby as they are favored by a few devs here).

I'm not saying, MS products should not be used. I'm just saying the state of open source products is not at all as you describe. Every company needs to go with what it's strengths are. It is obvious you are not strong in non-microsoft products. Just as I wouldn't know how to properly manage a MSSQL database or configure active directory. I could learn, but it would take 2 times the time and be riddled with problems as I learn on the job. I have never written in C# or even opened visual studio, if I was to write web apps using C# I'm sure it would be buggy and take 3 times as long as if you asked me to do it in languages I have written in for years.

This is not a flaw in MS products. This is a flaw in my knowledge. I have "big boy toys" to do real development with ruby, php, etc. I get enterprise level jobs done with these tools, and I get them done on time and on a shoestring budget. You seem to simply not be informed and instead just dismissive.

 

[PhatoseAlpha]

It's really kind of pointless to talk about the state of open source - it's an very broad category covering things ranging from well tested commerical grade programs to entirely useless crap, without anything unifying them save that they happen to give out source code. One might as well discuss whether apples in general were more rotten then oranges.

 

[Rangoric]

Originally posted by: degibson

Originally posted by: GodlessAstronomer

Originally posted by: brianmanahan
Sorry degibson, saying "M$" is so 1990's...

I stopped reading both of his posts the moment I saw that. Grow up.

Yes, please ignore the OP's question and comments about benchmarking, cost-effectiveness, etc., and focus on my choice of abbreviation. The internet is, after all, entertainment first and foremost.

It shows a strong bias. About on par with Linsux, Winblows, and other things of that nature.

And the bias isn't shown in your full post (besides that first paragraph) so you are doing yourself a disservice by using an abbreviation that you KNOW might cause problems.

 

[sourceninja]

Originally posted by: PhatoseAlpha
It's really kind of pointless to talk about the state of open source - it's an very broad category covering things ranging from well tested commerical grade programs to entirely useless crap, without anything unifying them save that they happen to give out source code. One might as well discuss whether apples in general were more rotten then oranges.

When I say open source, I refer to major well respected projects.

Mainstream linux distros, mainstream databases, mainstream programing languages. To say these are not mature, enterprise level tools is really a disservice.

 

[Macaw]

Maybe give Azure a look. I have used it with all my new projects and it is mighty difficult to go back to PHP/MySQL or hosted IIS/MSSQL after using Azure.

http://www.microsoft.com/windowsazure/

By using a cloud platform you can scale well, avoid difficult problems like sychronization, and not worry about hosting issues. Force.com, Google, Azure, and, to a lesser degree, Amazon all have platforms you can program against now.

It's pretty darn cool to spin up a new web server, application server, or database without doing any work and be able to deploy to it.

 

[Markbnj]

Originally posted by: Rangoric

Originally posted by: degibson

Originally posted by: GodlessAstronomer

Originally posted by: brianmanahan
Sorry degibson, saying "M$" is so 1990's...

I stopped reading both of his posts the moment I saw that. Grow up.

Yes, please ignore the OP's question and comments about benchmarking, cost-effectiveness, etc., and focus on my choice of abbreviation. The internet is, after all, entertainment first and foremost.

It shows a strong bias. About on par with Linsux, Winblows, and other things of that nature.

And the bias isn't shown in your full post (besides that first paragraph) so you are doing yourself a disservice by using an abbreviation that you KNOW might cause problems.

A lot of people use that abbreviation more or less reflexively, so as tempting as it is you can't always read into it as much as you want to. In any case, let's keep this discussion on track with respect to the OPs question. There are a lot of interesting points regarding the pros and cons of financing start-up development in different ways, without devolving into a debate on the elemental merits of FOSS vs. proprietary approaches.

 

[PhatoseAlpha]

The best advice I can give you on something like this is "Do not to take advice from anybody on the internet on the matter of OSS vs MS".

It's practically a holy war, and pretty much anyone who's going to spend the energy to reply is going to have already picked a side, and will likely defend it with religious fervor and not logic. If you want a decent comparison, you're going to have to do the research yourself.

 

[brandonb]

Originally posted by: sourceninja
Are you seriously trying to imply there are no good IDE's for ruby? That there is no easy ways to debug or test code written with ruby?

Not that I'm aware of. The guy at work who uses Ruby says there is no IDE comparable to Microsofts IDE... He uses Notepad, because he says everything else sucks. What IDE do you recommend for Ruby and I'll ask him if he has any experience with it.

 

[esun]

Originally posted by: brandonb

Originally posted by: sourceninja
Are you seriously trying to imply there are no good IDE's for ruby? That there is no easy ways to debug or test code written with ruby?

Not that I'm aware of. The guy at work who uses Ruby says there is no IDE comparable to Microsofts IDE... He uses Notepad, because he says everything else sucks. What IDE do you recommend for Ruby and I'll ask him if he has any experience with it.

If you're literally speaking of notepad.exe, the super-crappy text editor located in %SystemRoot%\system32\notepad.exe, then let me assure you it's not Ruby (or any associated dev tools/IDEs) that sucks, it's your coworker. Denigrating a technology or platform based on its worst users isn't really fair.

As for the OP, as always you have to balance your team's experience versus the cost of the technology. Many open source options may be cheaper in terms of licenses than MS products, but if your team needs months to get up to speed on those technologies, that will end up costing you more in the long run. I wouldn't worry about scalability as much since hardware is relatively cheap and most technologies will get you reasonably far unless you're doing something truly massive.

In short, use what you and your team know. That will be what ends up being cheapest in the long run.

 

[sourceninja]

Originally posted by: brandonb

Originally posted by: sourceninja
Are you seriously trying to imply there are no good IDE's for ruby? That there is no easy ways to debug or test code written with ruby?

Not that I'm aware of. The guy at work who uses Ruby says there is no IDE comparable to Microsofts IDE... He uses Notepad, because he says everything else sucks. What IDE do you recommend for Ruby and I'll ask him if he has any experience with it.

Off the top of my head:

Anjuta
netbeans
eclipse
xcode
textmate
coda
rubymine
radrails
Aptana - eclipse stand alone product
kDevelop

and then typing ruby IDE in google has more results then I could ever find the time to evaluate.

I have not used visual studio seriously after .NET was released. I had moved to linux by then and have bounced around there with different IDE's depending on my tasks. For web development I used a custom setup of vi for a long long time. When I started writting some desktop apps in python I moved to Anjuta. Later I did a lot of work with Eclipse because of my job. Now that I use a mac exclusively I bounce between 3 tools.

Xcode, textmate, and coda. I'm not a thick IDE kind of guy. I like some of the features of modern IDE's, but for the most part textmate fits all of my web development needs. Coda has been recently introduced to me by a intern and I'm starting to like it for web development. It combines 3 or 4 tools I use into a single well designed tool.

A side business I'm starting requires me to write a iphone application. This got me into using xcode. I have simply fallen in love with xcode and lately I've been using it for my web projects. I just love it. I've even written a few mac applications for personal use in it using applescript, ruby, and objective-C.

Like I said, it's about knowing what is out there, what skills you have in your department, and what makes sense with those two things known. Our department is full of linux users and solaris certified engineers. It wouldn't make sense for us to start running asp and mssql.

 

[zebano]

OSS vrs MS... use what you're developers know and want or be prepared to retrain / replace them.

 

[Atheus]

My advice is to ignore all the interesting options presented here and just go with PHP because it's tried and tested. The last thing you want is to start with some 'rapid' platform and find out halfway through it doesn't do something you want it to do.

 

[brandonb]

Originally posted by: sourceninja

Originally posted by: brandonb

Originally posted by: sourceninja
Are you seriously trying to imply there are no good IDE's for ruby? That there is no easy ways to debug or test code written with ruby?

Not that I'm aware of. The guy at work who uses Ruby says there is no IDE comparable to Microsofts IDE... He uses Notepad, because he says everything else sucks. What IDE do you recommend for Ruby and I'll ask him if he has any experience with it.

Off the top of my head:

Anjuta
netbeans
eclipse
xcode
textmate
coda
rubymine
radrails
Aptana - eclipse stand alone product
kDevelop

and then typing ruby IDE in google has more results then I could ever find the time to evaluate.

Thank you. I will run some of those by him and see if he has any experience with it. And I do agree with you, different projects and needs are better suited for some things other than others.

But like I mentioned, it was my experience (and most of it is pretty old) that open source did not work out well for me and my projects, but I was already used to MS products so it just came more natural to me. So, I'm sure that had alot to do with it.

 

[hooflung]

Here at work I use a combination of proprietary and open source tools.

Windows 2008 is our server environment. I use IIS 7 with various Microsoft official mods such as the Rewrite module. However, I use PHP as my scripting language. I currently run the native 64bit binary available from fusion-x lan. I use NetBeans IDE 6.7.1 here at work and a combination of NetBeans and PHPDesigner 2008 at home. I use CodeIgniter at work which is a very, very good, clean MVC framework.

At home I use Debian for my development. I have a complete MSDN and access to the entire Microsoft repository, so money is not the problem. I just prefer Debian in my own things because I tend to appreciate Ruby, Ruby on Rails and Mod_Rails ( Phusion Passenger ). I don't hold myself to one programming language. Windows usually is the last pedestrian on the street corner to get a latest release of a OSS language other than Python and Perl.

Now if I was on a Mac I'd be using Textmate like a boss. And I'd have the best of both worlds between *nix and Microsoft. But I'd still be targeting Linux ( or Solaris/FreeBSD ) and Windows.

As for a startup the best way to go is with going with what you know. Maybe you want to do something similar to what I do at work. Use Microsoft Windows Server 2008 as the base then use PHP ( you can use the latest 32bit 5.3.0 which adds namespaces ) in FastCGI and get familiar with CodeIgniter. Use MySQL 5.0 or Postgresql 8 for your database. Alternatively, you can use Debian ( or Nexenta ) and run a Ruby on Rails shop. This will require ALL of you to get a level playing field, not just your Windows-centric partners.

Or divvy up the workload and go with the platform that the person doing the majority of the coding knows. If that is not you, then let them do VB/VC/.NET and you work on smaller tasks and maybe pick up more marketing and client relations ( you can actually talk on the technical prowess of your product even if you don't code most of it ).

Those are some ideas. But make sure whatever you and your partners don't burn out before you have a working product.