We're trying to figure out if a certain type of web server architecture is better than our current web server. We need to A/B test these 2 boxes with live traffic to really get a sense of which one performs better on our internal KPIs. The hosts involved can't help with the A/B test. So naturally I'm looking to see if it's possible to roughly split the traffic 50:50 between the servers using PHP redirects and sub domains. For example:
- www.mydomain.com - (randomly redirects all requests to 1 of the following sub domains, maintaining url structure and query strings...)
- www1.mydomain.com (original webserver config)
- www2.mydomain.com (new webserver config)
We would need to run this test for at least a week or two to generate good stats. What I'm not sure about is whether the search engines would see this as bad and effect our rankings.