Bufferbloat

From Wikipedia, the free encyclopedia
Jump to navigation Jump to search

Bufferbloat is a cause of high latency in packet-switched networks caused by excess buffering of packets. Bufferbloat can also cause packet delay variation (also known as jitter), as well as reduce the overall network throughput. When a router or switch is configured to use excessively large buffers, even very high-speed networks can become practically unusable for many interactive applications like voice over IP (VoIP), online gaming, and even ordinary web surfing.

Some communications equipment manufacturers designed unnecessarily large buffers into some of their network products. In such equipment, bufferbloat occurs when a network link becomes congested, causing packets to become queued for long periods in these oversized buffers. In a first-in first-out queuing system, overly large buffers result in longer queues and higher latency, and do not improve network throughput.

The bufferbloat phenomenon was initially described as far back as in 1985.[1] It gained more widespread attention starting in 2009.[2]

Buffering[edit]

An established rule of thumb for the network equipment manufacturers was to provide buffers large enough to accommodate at least 250 ms of buffering for a stream of traffic passing through a device. For example, a router's Gigabit Ethernet interface would require a relatively large 32 MB buffer.[3] Such sizing of the buffers can lead to failure of the TCP congestion control algorithm. The buffers then take some time to drain, before congestion control resets and the TCP connection ramps back up to speed and fills the buffers again.[4] Bufferbloat thus causes problems such as high and variable latency, and choking network bottlenecks for all other flows as the buffer becomes full of the packets of one TCP stream and other packets are then dropped.[5]

A bloated buffer has an effect only when this buffer is actually used. In other words, oversized buffers have a damaging effect only when the link they buffer becomes a bottleneck. The size of the buffer serving a bottleneck can be measured using the ping utility provided by most operating systems. First, the other host should be pinged continuously; then, a several-seconds-long download from it should be started and stopped a few times. By design, the TCP congestion avoidance algorithm will rapidly fill up the bottleneck on the route. If downloading (and uploading, respectively) correlates with a direct and important increase of the round trip time reported by ping, then it demonstrates that the buffer of the current bottleneck in the download (and upload, respectively) direction is bloated. Since the increase of the round trip time is caused by the buffer on the bottleneck, the maximum increase gives a rough estimation of its size in milliseconds.[6]

In the previous example, using an advanced traceroute tool instead of the simple pinging (for example, MTR) will not only demonstrate the existence of a bloated buffer on the bottleneck, but will also pinpoint its location in the network. Traceroute achieves this by displaying the route (path) and measuring transit delays of packets across the network. The history of the route is recorded as round-trip times of the packets received from each successive host (remote node) in the route (path).[7]

Mechanism[edit]

Most TCP congestion control algorithms rely on measuring the occurrence of packet drops to determine the available bandwidth between two ends of a connection. The algorithms speed up the data transfer until packets start to drop, then slow down the transmission rate. Ideally, they keep adjusting the transmission rate until it reaches an equilibrium speed of the link. So that the algorithms can select a suitable transfer speed, the feedback about packet drops must occur in a timely manner. With a large buffer that has been filled, the packets will arrive at their destination, but with a higher latency. The packets were not dropped, so TCP does not slow down once the uplink has been saturated, further filling the buffer. Newly arriving packets are dropped only when the buffer is fully saturated. Once this happens TCP may even decide that the path of the connection has changed, and again go into the more aggressive search for a new operating point.[8]

Packets are queued within a network buffer before being transmitted; in problematic situations, packets are dropped only if the buffer is full. On older routers, buffers were fairly small so they filled quickly and therefore packets began to drop shortly after the link became saturated, so the TCP protocol could adjust and the issue would not become apparent. On newer routers, buffers have become large enough to hold several megabytes of data, which translates to time amounts in seconds required for emptying the buffers. This causes the TCP algorithm that shares bandwidth on a link to react very slowly as its behavior depends on actually having packets dropped when the transmission channel becomes saturated.

The problem also affects other protocols. All packets passing through a simple buffer implemented as a single queue will experience the same delay, so the latency of any connection that passes through a filled buffer will be affected. Available channel bandwidth can also end up being unused, as some fast destinations may not be reached due to buffers clogged with data awaiting delivery to slow destinations — caused by contention between simultaneous transmissions competing for some space in an already full buffer. This also reduces the interactivity of applications using other network protocols, including UDP or any other datagram protocol used in latency-sensitive applications like VoIP and games.[9] In extreme cases, bufferbloat may cause failures in essential protocols such as DNS.

Impact on applications[edit]

Any type of a service which requires consistently low latency or jitter-free transmission (whether in low or high traffic bandwidths) can be severely affected, or even rendered unusable by the effects of bufferbloat. Examples are voice calls, online gaming, video chat, and other interactive applications such as instant messaging and remote login. Latency has been identified as more important than raw bandwidth for many years.[citation needed]

When the bufferbloat phenomenon is present and the network is under load, even normal web page loads can take many seconds to complete, or simple DNS queries can fail due to timeouts.[10]

Diagnostic tools[edit]

The DSL Reports Speedtest[11] is an easy-to-use test that includes a score for bufferbloat. The ICSI Netalyzr[12] is another on-line tool that can be used for checking networks for the presence of bufferbloat, together with checking for many other common configuration problems.[13] The bufferbloat.net web site also provides an easy procedure for determining whether a connection has excess buffering that will slow it down.[14]

Solutions and mitigations[edit]

Several technical solutions exist which can be broadly grouped into two categories: solutions that target the network and solutions that target the endpoints. These types of solutions are often complementary.

Network solutions generally take the form of queue management algorithms. This type of solution has been the focus of the IETF AQM working group[15]. Notable examples include:

  • AQM algorithms such as CoDel and PIE.[16]
  • Hybrid AQM and packet scheduling algorithms such as FQ-CoDel.[17]
  • Amendments to the DOCSIS standard[18] to enable smarter buffer control in cable modems.[10]
  • Integration of queue management into the WiFi subsystem of the Linux operating system.[19]

Notable examples of solutions targeting the endpoints are:

  • The BBR congestion control algorithm for TCP.
  • The µTP protocol employed by many BitTorrent clients.
  • Techniques for using fewer connections, such as HTTP pipelining or HTTP/2 instead of the plain HTTP protocol.[10]

For most end-users, the largest improvement can be had by fixing their home router[20]. Most of the technical fixes are included in recent versions of the Linux operating system and the LEDE aftermarket router firmware.[20] The problem may also be mitigated by reducing the buffer size on the OS[10] and network hardware; however, this is often not configurable.

See also[edit]

References[edit]

  1. ^ "On Packet Switches With Infinite Storage". 1985-12-31.
  2. ^ van Beijnum, Iljitsch (2011-01-07). "Understanding Bufferbloat and the Network Buffer Arms Race". Ars Technica. Retrieved 2011-11-12.
  3. ^ Guido Appenzeller; Isaac Keslassy; Nick McKeown (2004). "Sizing Router Buffers" (PDF). ACM SIGCOMM. ACM. Retrieved 2013-10-15.
  4. ^ Nichols, Kathleen; Jacobson, Van (2012-05-06). "Controlling Queue Delay". ACM Queue. ACM Publishing. Retrieved 2013-09-27.
  5. ^ Gettys, Jim (May–June 2011). "Bufferbloat: Dark Buffers in the Internet". IEEE Internet Computing. IEEE. pp. 95–96. doi:10.1109/MIC.2011.56. Retrieved 2012-02-20.
  6. ^ Clunis, Andrew (2013-01-22). "Bufferbloat demystified". Retrieved 2013-09-27.[self-published source?]
  7. ^ "traceroute(8) – Linux man page". die.net. Retrieved 2013-09-27.
  8. ^ Jacobson, Van; Karels, MJ (1988). "Congestion avoidance and control" (PDF). ACM SIGCOMM Computer Communication Review. 18 (4). Archived from the original (PDF) on 2004-06-22.
  9. ^ "Technical Introduction to Bufferbloat". Bufferbloat.net. Retrieved 2013-09-27.
  10. ^ a b c d Gettys, Jim; Nichols, Kathleen (January 2012). "Bufferbloat: Dark Buffers in the Internet". Communications of the ACM. 55 (1). ACM: 57–65. doi:10.1145/2063176.2063196. Retrieved 2012-02-28.
  11. ^ "Speed test - how fast is your internet?". dslreports.com. Retrieved 26 October 2017.
  12. ^ "ICSI Netalyzr". berkeley.edu. Retrieved 30 January 2015.
  13. ^ "Understanding your Netalyzr results". Retrieved 26 October 2017.
  14. ^ "Tests for Bufferbloat". bufferbloat.net. Retrieved 26 October 2017.
  15. ^ "IETF AQM working group". ietf.org. Retrieved 26 October 2017.
  16. ^ Pan, Rong; Natarajan, Preethi; Piglione, Chiara; Prabhu, Mythili; Subramanian, Vijay; Baker, Fred; VerSteeg, Bill (2013). PIE: A Lightweight Control Scheme To Address the Bufferbloat Problem. 2013 IEEE 14th International Conference on High Performance Switching and Routing (HPSR). IEEE. doi:10.1109/HPSR.2013.6602305.
  17. ^ Høiland-Jørgensen, Toke; McKenney, Paul; Taht, Dave; Gettys, Jim; Dumazet, Eric (2016-03-18). "The FlowQueue-CoDel Packet Scheduler and Active Queue Management Algorithm". Retrieved 2017-09-28.
  18. ^ "DOCSIS "Upstream Buffer Control" feature". CableLabs. pp. 554–556. Retrieved 2012-08-09.
  19. ^ Høiland-Jørgensen, Toke; Kazior, Michał; Täht, Dave; Hurtig, Per; Brunstrom, Anna (2017). Ending the Anomaly: Achieving Low Latency and Airtime Fairness in WiFi. 2017 USENIX Annual Technical Conference (USENIX ATC 17). USENIX - The Advanced Computing Systems Association. pp. 139–151. ISBN 978-1-931971-38-6. Retrieved 2017-09-28.
  20. ^ a b "What Can I Do About Bufferbloat?". bufferbloat.net. Retrieved 26 October 2017.

External links[edit]