class of service algorithms class based weighted fair queuing cbwfq modified deficit round robin mdrr


Class of Service: Myths and Misconceptions
Multinational enterprises are experiencing relentless pressures from inside and outside their organizations fueled by “the perfect storm” CoS Does Not Change the Laws of Physics of multiple forces colliding – application development, technology advancements, more savvy and demanding customers and the There are four sources of packet delay in a typical enterprise environment – forwarding delay1, queuing delay, propagation delay2 and serialization delay3. Assuming that the enterprise has deployed economics of an increasingly competitive marketplace. Utilizing Class of Service (CoS) functionality is one method of optimizing network the industry standard for hardware-based routers and the access bandwidth is fixed, you have no real control over forwarding delay,
performance so that it aligns with the enterprise agenda. propagation delay or serialization delay. The only aspect of packet delay across the network that can be actually controlled is queuing There has been much discussion about the valuable benefits of CoS; however, there are also some common myths, which unless delay. Queuing delay (1/1-U) is best observed by understanding the load across the network in both directions. This utilization will challenged, can impact the success of the CoS implementation. The focus of this paper is to expose the common misconceptions increase the overall time a packet has to wait in transit for the line to clear. Load on the network can quickly start to increase packet delay; of what CoS can and cannot do. We will also discuss an approach to ensure a successful CoS implementation in the enterprise – and
70% (for tame “the perfect storm.” Myths and Misconceptions Figure 1: Serialization Delay Increase as Function of Utilization CoS Adds Bandwidth and Relieves Congestion CoS is a powerful network design tool to aid in performance engineering for your network. Used properly, these features often Effect of Utilization on Queuing Delay allow you to forgo bandwidth upgrades while maintaining the performance of mission critical applications. However, CoS is not a 15 substitute for insufficient bandwidth, nor does it replace the need for careful capacity planning. CoS is intended to provide deterministic Delay Factor 10 behavior during periods of access facility congestion. This behavior represents a tradeoff usually favoring mission critical applications
during congestion over less time-sensitive or less critical applications. If sustained congestion occurs for a given network link, then additional capacity may be the only way toassure satisfactory performance for 0
the applications supported across the connection. Utilization Figure 1: Serialization Delay Increase as Function of Utilization Without CoS, queuing delay can easily represent the largest delay component across the network. When CoS is properly deployed, the Working closely with the service provider, IT managers can still develop a very granular approach to traffic classification and prioritization. A queuing delay is minimized and the end-to-end delay is primarily a function of the fixed delay components. The deployment of CoS in the service provider need not offer the same number of classes that are defined in the enterprise. QoS transparency allows enterprises to network does not impact the speed of light in optical fiber, the switching time of networking equipment or the time taken to transmit a large packet onto a low speed link. Therefore, even with a proper retain their DSCP code markings4 such that traffic going through the service provider network will translate back to the finer-grained class scheme of the local enterprise network.
CoS strategy, any of these remaining delay components can still lead to marginal performance of response-time critical applications. The More Classes of Service You Have – The Better, Right? CoS Controls All Aspects of the End-User Experience One of the main tools available for implementing CoS is the intelligent scheduling of packets. Queuing algorithms are therefore crucial The end-user’s experience is controlled by factors other than CoS. Some applications perform poorly across WAN facilities even with components in effective congestion control as they determine the way in which packets from different sources interact with each other, which optimized CoS. In fact, some applications will perform poorly when there is no competing traffic on the WAN. These applications are in turn affects the collective behavior of flow control algorithms. It is the scheduler that chooses in what order service requests are allowed often performance bound by application level windowing and/or back-end processing that cannot be overcome by CoS optimization. access to resources; its service discipline dictates how to multiplex packets from different connections and decides which packets to
Microsoft Networking, native database operations and many file transfer utilities are a few of the examples of applications whose transmit. The three most important resources allocated by the service discipline are bandwidth (packet volume), promptness (when packets inherent implementation limits network performance. are transmitted) and buffer space (which packets are discarded). Additional factors can skew response time results including performance of user workstations, performance of the DNS server, protocol windowing, stability of the LAN environment and performance There is a common misconception that the more levels of CoS, the better the network. In reality, however, an overly complicated CoS of the target application servers. policy is an operational hazard. The more CoS you have, the more expensive it becomes in terms of classification, queuing, provisioning Every Application Needs its Own Class and management. While it is theoretically possible to ensure perfect separation by using per-flow queues, the overhead associated with At the outset of the network design, it is necessary to identify the different characteristics of the traffic on the network. Voice traffic such extreme scheduling leads to unacceptable performance under congested conditions. typically requires specific latency, jitter and bandwidth characteristics. ERP is less susceptible to jitter, but still requires low latency to prevent The reality is that, based on an analysis of the statistical nature of IP traffic and the way this impacts the performance of voice, video and the session from timing out, and typical bandwidth of 25Kbps per user. In comparison, Microsoft Exchange database replication has no strict data services, performance requirements are satisfied without explicit service differentiation of every application, thus creating a relatively latency or jitter requirement, but will consume as much bandwidth as is available, often to the detriment of other applications. Beyond simple CoS platform for the converged network. these three typical applications lie a variety of needs from both business-critical to less critical applications. On the surface, it seems Conventional wisdom now suggests that 4-6 classes of service are the optimal number, configured as follows: voice, video, mission-critical relatively easy to apportion service classes to applications. The most critical, lowest and consistent latency service class should go to voice, transactional data, normal business traffic and bulk data. This range may be considered the optimal number for several reasons. First, from the next most critical class should go to ERP or some other client- server business application and the least critical, best-effort service a business perspective, 4-6 classes of service are sufficient to classify traffic on the basis of its “value” to the business and its technical class should go toall non-latency-sensitive applications such as Exchange Database replication. performance requirements. Voice and video require different types of prioritization (voice is very sensitive to jitter, video is less so) and data
While it is certainly tempting to define a separate class for every application, there reaches a point of diminishing returns in terms of can almost always be classified properly into three classes. Finally, 4-6 classes provide reasonable scheduling behavior in most routing platforms and on the most predominant WAN link speeds.
CPE resources and application performance. The reality is that in practice, optimal application performance is obtained by grouping CoS Strategies are Universal Regardless of Port Speed applications with similar performance requirements together in the same class. Another real and significant problem for the enterprise is Port speed has a significant impact on CoS behavior. In other words, there is no universal CoS strategy with regard to port speed. The the inability toaccurately identify every application on the network – mainly due tolack of visibility intothe network. This can lead to the discernment between high speed (port sizes >T1) and low speed (port sizes Common scheduling algorithms in use today such as Class-Based Weighted Fair Queuing (CBWFQ) and Modified Deficit Round Robin network is ever-changing. However, adhering to the strategies and avoiding the myths that were discussed in this paper will help move (MDRR) serve the traffic in the various classes based on a ‘bandwidth’ assignment. A higher bandwidth assignment typically results in more the IT manager a step closer to WAN optimization across the converged voice/video/data network infrastructure. frequent servicing of the class. For high speed environments, the typical approach should be to align the class allocations as closely as possible to the expected application Network Design and Consulting Division Helping customers build an optimal CoS design is one of many services provided by the consultants and analysts of ’s Network Design volumes for the classes. However, this approach can be problematic for ‘low-speed’ links. and Consulting Division. Based on years of experience supporting thousands of customers, this team leverages state-of-the-art design Response time critical applications tend to consume the least bandwidth, and as such, would intuitively justify a low bandwidth allocation for the methodologies and tools to help customers meet their business requirements through best-in-class network solutions. class. On low speed links, the infrequent servicing of low bandwidth classes can lead to poor/erratic performance of interactive applications, Notes
1. Forwarding Delay (also known as switching delay) is the time it takes for a router or switch to process the bits and forward/route/switch even when the allocation is sufficient from a bandwidth perspective. To counter this behavior, response time critical classes are typically ‘over-allocated’ for low speed links. The result is that the response time classes are checked for servicing frequently, keeping delay to a minimum. them onto the next interface. It does NOT include the delay at the client or at the server, but clearly any slowdowns at these components will affect the packets that are being generated/received between them. And since the class does not really consume this over-allocation, the left over bandwidth is still available to the remaining classes.
2. Propagation delay is what most people think of when they use the term delay – it is the absolute speed of Electromagnetic energy Conclusion The past several years have witnessed a dramatic increase in the number and variety of applications running over the enterprise through a medium. Most network professionals use 10msec/1000 miles to account for retransmission/signal regenerator as well as any route inefficiencies that exist. network. This has certainly driven the need to support different levels of CoS for these diverse application types. Realization of CoS
3. Serialization delay is a function of the packet size divided by the access line rate. Example – it takes 187.5 msec to insert a 1500 byte deployment requires association/mapping of the different applications with CoS markings, the determination of the CoS to be provided to packet onto a 64 Kbps line – do this twice (once in and once out) and we have an average insertion delay component of 375 msec! each and finally, the mechanisms in the underlying network to deliver CoS. CoS mapping and deployment is a non-trivial task which has
4. Differentiated Services Code Point RFC 2474. proven to be all the more challenging to the IT manager since the
In cities, towns and remote locations, mini satellite dishes point attentively to the Southern sky. Emblazoned with names like DirecTv, ExpressVu, DirecWay, Web Conferencing, iNetVu, Linkstar, XM Satellite Radio, Sirius Satellite Radio iDirect their numbers are growing at an amazing rate. Iridium Satellite Phone is the only provider of truly global satellite voice and data solutions with complete coverage of the earth (including oceans, airways and Polar Regions). Get the latest buzz on Free satellite tv systems - including the features and benefits that make them today's ultimate television viewing experience.(Get Dish) Ever wonder why these satellite dish systems are in such great demand? Does high speed internet service or digital television programming via satellite intrigue you?
If you've never heard of Cheap VoIP, get ready to change the way you think about long-distance phone calls. VoIP, or Voice over Internet Protocol, is a method for taking analog audio signals and turning them into digital data (IP packets) that can be transmitted over the Internet.