All Products
Document Center

Billing Method

Last Updated: Mar 21, 2018

The billing structure of Dynamic Route for CDN consists of basic service and quantity of requests.

For information on the exact amount billed, see Dynamic Route for CDN Pricing Details. The billing methods are described as follows.

Basic service

  • Two billing method options are available: PayByBandwidth or PayByTraffic.
  • You can flexibly switch between these two billing methods. Following a switch, the new billing method takes effect for the next billing cycle.

Applicable scenarios for the two billing methods are as follows:

Billing method Description Applicable scenarios
PayByBandwidth Users are billed by daily peak bandwidth
. A peak bandwidth value is recorded at intervals of 5 minutes for a total of 288 values each day. The highest value is used for billing.
Where the domain traffic curve is relatively stable
or where the daily bandwidth utilization rate is greater than 30%
PayByTraffic Billed by actual daily traffic Where large fluctuations exist in the domain traffic curve and with bandwidth spikes
or where the daily bandwidth utilization rate is less than 30%

Bandwidth utilization rate = Used traffic GB/(peak bandwidth (Mbps) x 10.54). 100% use of 1 Mbps bandwidth generates approximately 10.54 GB traffic every day.

Quantity of requests

The total number of Dynamic Route for CDN requests is comprised of the following three items, which are billed according to their unit prices on a daily basis:

  • Quantity of dynamic HTTP requests
  • Quantity of dynamic HTTPS requests
  • Quantity of static HTTPS requests

Additional information

Metering data statistics

The system collects summary data from the domain names of the nodes across the Dynamic Route for CDN at a time granularity of 5 minutes (288 nodes a day). This data is used as the basis for billing.

Differences between billed traffic and log-monitored traffic

Traffic data recorded in the Dynamic Route for CDN domain log is counted by the application layer log. However, actual network traffic is 7-15% higher than the traffic counted by the application layer. This is largely due to two reasons:

1. Consumption by TCP/IP packet header: HTTP requests are widely known to be based on TCP/IP. On the Internet, each packet contains up to 1,500 bytes, including a 40-byte header inserted by TCP/IP. This packet header also generates traffic. However, the act of adding the packet header is carried out by the core protocol stack and cannot be counted by the application layer. As such, the log does not record these 40 bytes. This portion of traffic accounts for more than 2.74% (40/1460) of the log-recorded traffic. Typically, it is around 3%.

2. TCP retransmission: Depending on the physical Internet load, about 3-10% of the packets we send are discarded by the Internet. The server then re-transmits the discarded packets over the core protocol stack, but the application layer cannot count these packets. This traffic may make up a percentage of the logged traffic, which may vary depending on the Internet connection quality. In the early hours of the day when Internet traffic is light, the re-transmission rate is low. During peak hours, such as in the evening, the rate may rise. Typically, it is between 3-7%.

Therefore, based on industry standards, an additional 7-15% Internet traffic overhead is included in chargeable traffic. Alibaba Cloud Dynamic Route for CDN takes the average of 10% and adds it to the statistics for Internet traffic consumption.