Increasing the number of points of failure absolutely does increase the chances of failure, even if each 'point' is more reliable. 1 server with a 98% uptime is more reliable than 5 servers with a 99% uptime, if all 5 have to be working for everything to 'work'.