Home ›

Kernel Korner - Analysis of the HTB Queuing Discipline

Issue 131

From Issue #131
March 2005

Jan 26, 2005  By Yaron Benita
 in
Can Linux do Quality of Service in a way that both offers high throughput and does not exceed the defined bandwidth? Here's a thorough test.
Testing Model 2

Figure 10. Test 1: Three streams In, one stream Out

Table 8. Test 1 Results

StreamBurst (Bytes)Cburst (Bytes)Packet-Size (Bytes)In-Bandwidth (Mbit/s)Out-Bandwidth (Mbit/s)Class
11k1k51250.6502
21k1k51250.6002
31k1k51250.5682
Total18k18k–151.818–

As shown in Table 8, the bandwidth is distributed evenly when several streams are transmitting the same number of bytes and belong to the same class. Another test, in which the input bandwidth in stream 1 was higher than that of streams 2 and 3, showed that the output bandwidth of stream 1 also was higher than streams 2 and 3. From these results, it can be concluded that if more data is sent on a specific stream, the stream is able to forward more packets than other streams within the same class.

Conclusions

The test cases presented here demonstrate one way to evaluate HTB accuracy and performance. Although continuous packet bursts at a specific rate don't necessarily simulate real-world traffic, it does provide basic guidelines for defining the HTB classes and their associated attributes.

The following statements summarize the test case results:

  • The maximum bandwidth that a Linux machine can forward (receive on one interface and transmit on another interface) with continuous streams of 64-byte packets is approximately 34Mbit/s. This upper limit occurs because every packet that the Ethernet driver receives or transmits generates an interrupt. Interrupt handling occupies CPU time, and thus prevents other processes in the system from operating properly.

  • When setting the traffic rate to 30Mbit/s, the cburst/burst values, which give the most average accuracy results, are 18k/18k.

  • There is a linear relationship between the burst value and the requested rate. The cburst/burst values of a 30Mbit/s rate can be used as a starting point for calculating the burst values for other rates.

  • It is possible to control the throughput accuracy by changing the cburst/burst values. The accuracy bandwidth range size is approximately 2Mbit/s for 64–1,500 byte packet sizes.

  • Bandwidth is distributed evenly when several streams are transmitting the same number of bytes and belong to the same class.

Resources for this article: www.linuxjournal.com/article/7970.

Yaron Benita is originally from Jerusalem, Israel, and currently lives in San Francisco, California. He is the CMTS software manager at Prediwave. He works mostly in the networking and embedded fields. He is married and has a lovely six-month-old daughter. He can be reached at yaronb@prediwave.com or ybenita@yahoo.com.

______________________

Webcast
More Resources
How to Build an Optimal Hadoop Cluster to Store and Maintain Unlimited Amounts of Data Using Microservers

Realizing the promise of Apache® Hadoop® requires the effective deployment of compute, memory, storage and networking to achieve optimal results. With its flexibility and multitude of options, it is easy to over or under provision the server infrastructure, resulting in poor performance and high TCO. Join us for an in depth, technical discussion with industry experts from leading Hadoop and server companies who will provide insights into the key considerations for designing and deploying an optimal Hadoop cluster.

Learn More

Sponsored by AMD

White Paper
More Resources
Red Hat White Paper: Using an Open Source Framework to Catch the Bad Guy

Built-in forensics, incident response, and security with Red Hat Enterprise Linux 6

Every security policy provides guidance and requirements for ensuring adequate protection of information and data, as well as high-level technical and administrative security requirements for a system in a given environment. Traditionally, providing security for a system focuses on the confidentiality of the information on it. However, protecting the data integrity and system and data availability is just as important. For example, when processing United States intelligence information, there are three attributes that require protection: confidentiality, integrity, and availability.

Learn more about catching the bad guy in this free white paper.

Learn More

Sponsored by DLT Solutions