Developing a large scale, heavy traffic site.

[epreston]

I think I pretty much know what to expect for an answer, but I still have to ask...

If, say, facebook or craigslist were only in the idea stage (not back in the day but now) could either of them be built more rapidly, smoothly and reliably upon a framework like processwire than custom-from-the-ground-up? And could the performance (traffic & bazillions of simultaneous users) of the end result be as speedy as a custom built effort?

I'm certain it could get built more rapidly, but is there any price to pay on the other end if/when performance becomes an issue?

Thanks,

Ed

[ryan]

Could it be built more rapidly, smoothly and reliably with ProcessWire? Absolutely. Could the same be said of using a quality framework vs. starting from scratch? Yes, though I'd personally choose ProcessWire. Would it be as scalable as building something from scratch, and optimizing it specific to the need? No way. As Facebook grew, they even went as far as to build their own PHP, separate from the one you can I use. Using an existing PHP framework like ProcessWire is an excellent way to take a potentially big idea to fruition. If the idea is a good one that delivers on being the next Facebook or Craigslist, then you'll have no problem getting the resources to take your application code where you need to. I think most of the heavy-hitting services out there start this way, and then move on to something custom once the scale justifies it. But that scale is extremely large. It would be cost prohibitive to build for that scale when you haven't yet reached it. 

In terms of price to pay, the most likely needs for scale can be solved by increasing your server's resources or switching to a cloud-based hosting environment that can scale larger than an individual server. But again, if things get Facebook-large, then you'll be in an entirely different segment where you'll probably be building not just your own software, but your own hardware too.  

[epreston]

That more than answers my question, and in fact is exactly what I was hoping to hear!

And thanks for that bit of info about Facebook developing their own PHP - very interesting! I could tell there was some serious custom engineering behind those Facebook features, but I had no idea they want that far.

[SiNNuT]

If you do not take scale into account i think you could probably build some of the functionality pretty fast and efficiently in PW, but in all honesty, if you where to build the next Facebook you will be looking for other tools.

If you take it to Facebook scale there is no way you would do it PW. Custom is the way to go here. That's not a knock on PW though. Heck, next to Hiphop php, those FB guys are building their own storage engines, webservers etc. http://www.slideshare.net/mozion/facebook-architecture-for-600m-users

[apeisa]

You could definitely use PW to build Facebook (feature wise). There is absolutely no reason to worry about being biggest and most active website of all, before you haven't written a line of code for the project.

I have build websites for over 15 years (8 of them professionally) and there have been very few scalability issues. But oh boy I have worried. The very same question that epreston asked: does it scale?

After these 15 years I am pretty sure it is much harder to build service that will be so popular that actually needs to be scaled - than the actual scaling.