What kind of hosting web server / DB / infrastructure is need for high volume dating website?

Question

I want to know what kind of infrastructure is need for a dating website developed in php/mysql. Consider a website like match.com and eharmony.com with millions of users. Users create their profile, can upload several photos, serach profiles and chat and Emails.

So what kind of hosting plan is needed for such website? How much DiskSpace is ideal? Any limitations with Database? What kind of Server Configuration?

What kind of plan/server configuration should I start with initially? say my site is going to be about one-third of match.com

When should I move to a higher level plan?

Answer 1

I found this document which describes the architecture of match.com http://www.plentyoffish.com/match.pdf. At one point they had 120 servers. So you're looking at getting 40 servers. You'll want them all to yourself so you need to pay for dedicated hosting, or buy them yourself and rent space in a data centre.

You need to move to a higher level when your servers are beginning to get overloaded.

How much disk space - it depends how much data you have.

Server configuration is something that can't be easily answered in a question like this, but you'll probably want a front end server that passes out requests to other servers. You'll want to separate out the databases from the servers doing scripting and the servers serving up static content.

To start off though, you could have two servers, one for the database, one for everything else.

This question isn't quite the same as yours, but has some pointers to useful resources: https://serverfault.com/questions/84715/nix-cloud-cluster-solutions-for-scalable-web-services

Answer 2

Having "platform templates" is a good start. I would recommend contacting "cloud host providers" Talk to their design architects, with those parameters. I assume you have growth prospects? You have many of the questions to ask. They should know what other questions to ask.

The whole theory of cloud computing is based on highly flexible numbers of CPU cores, RAM per device, Storage needed. Availability requirement. Ease of scalability. Costs per "unit"

Otherwise, the capital expenses of DataCenter specifications, infrastructure, cooling, scalability, staffing, benefits, security, network throughput, redundancy, (I can go on & on) all must be considered.

Having built and designed hundreds of networks, in place all over the world. All this during my 28+ year carrier gives me experience building as such for a World 5 corporation, Universities, Academic Not for Profits, Small Corporations.