The embodiments are directed to methods and apparatuses for pairing delivery group machines with one or more server farms in computing networks. The methods and apparatus can allocate machines in a delivery group across one or more server farms in a manner that maximizes efficiency through better computer resource usage. The methods and apparatuses select a server farm having a largest available capacity, and allocate machines from a delivery group to the server farm. If the quantity of delivery group machines exceeds the server farm capacity, the remaining machines are allocated to a second server farm. The methods and appliances also provide for de-allocating server farms, by selecting a server farm with the least allocated capacity, and de-allocating one or more delivery group machines from the selected server farm to reduce the number of utilized server farms.