We charge based on Total Transfer billing which suits Managed Hosting customers just fine. Why? It results in lower bandwidth charges.
Managed Hosting customers generally experience sustained levels of traffic over their billing month as oppose to traffic with sudden peaks. That's why billing based on the commonly used 95th Percentile (or burstable billing) method is not ideal, as it would result in paying for bandwidth that isn't used during the month. Total Transfer billing ignores these natural peaks and valleys, and only measures your actual data to and from your IP address. As a result, we do not use port based monitoring to determine your bandwidth traffic because the traffic crossing your public port would count against your bandwidth allocation. Instead, we use network-wide monitoring which measures the traffic flow on our entire network. Again, only the traffic to and from your IP address is counted towards your bandwidth allocation. Nothing else.
How Does it Work
We take a sampling of your traffic coming in and out of our datacenter at the edge of our FastFiber Network™. The sample ignores any traffic sitting on the network and only counts the traffic between your servers and your customers. This data is presented as a bandwidth graph inside the MyPEER1 customer portal where we group your servers into Bandwidth Allocation Groups (BAGs). A BAG includes a single server with a single IP address or a single server with multiple IPs. For more complex hosting environments, multiple servers and their IPs are also be combined into a BAG. We measure the bandwidth usage of each BAG by pooling all IPs assigned to the server. You then have the choice of viewing your bandwidth usage by BAG, server, or individual IP address in daily, weekly, or monthly intervals.
Monitor Your Usage in Simple Graphs
Your bandwidth usage is displayed in three formats - a line graph and two separate bar charts - with blue for inbound traffic and green for outbound traffic. The line graph (A) shows your bandwidth usage in daily bytes. Move your cursor over an area within the graph and you'll see direction, time, and amount. The first bar chart (B) shows your total BAG bandwidth usage in GB, with each section within the bar displayed proportional to the total. The second bar chart () shows your total BAG usage in GB compared to your allocated amount. If you need a more granular view, expand each BAG to list all the servers within it, and each server to list all the IPs assigned to it. Measure each at any level for a given time period.
Bill Back Your Customers for Bandwidth
Our Bandwidth Sampling Formula
A number of calculations are then performed to interpolate over the excluded frames and arrive at a very accurate estimate for what the actual usage is. The more times an IP's traffic is sampled, the more accurate the results. The margin for error drops to less than 0.073% at 10TB. Even at a few hundred gigabytes, the error is still less than 5%.
Here is the simplified equation for estimating the error.
c is the number of samples of a particular class we have.
%error <= 196 * sqrt(1 / c)
Here's a chart of what this equation looks like:
We have configured our system to include all the IP addresses that are under our management. This means we know without a doubt if an IP is foreign and does not reside on our network. When we retrieve data for bandwidth billing, we include only inbound/outbound and outbound/inbound traffic. Internal traffic is excluded and is not counted in billing. For this reason, internal traffic never shows up on the customer portal.