ADVANCED QUALITY OF SERVICE (QoS)
Welcome to the world of Quality of Service (QoS) and Quality of Experience (QoE) where we ensure that your mission-critical and important data applications get supreme priority over the other types of traffic and as it passes through the network to avoid congestion and bottleneck areas of your network. It is also here where we ensure that your VoIP and Video over IP gets the necessary priority they need due to their delay-sensitive nature.
Our job is to ensure that we assess the applications that are running in your network and then design the appropriate QoS strategy for you. We well know that no two different networks are the same and that is why a QoS solution that works for a given customer does not necessarily have to work the next customer and that is why we do due diligence to ensure we get each customer’s specific needs. Talk to us and see what we can do for you.
QoS is a fundamental network infrastructure technology—in the same class as high-availability and security technologies. Like these other technologies, the basics of QoS have remained fairly steady for many years now, but there has always been a continuing evolution in the refinement and sophistication of specific QoS mechanisms and in the breadth of platforms where these tools are available. In addition, network and user requirements, and application types and volume, have all changed dramatically in the past few years and are continuing to do so.
QOS DESIGN OVERVIEW
For a successful QoS deployment, we use these tools, which include classification and marking tools, policing and shaping tools, queuing and dropping tools, bandwidth-reservation tools, and advanced tools like Medianet and application visibility and control.
The fundamental purpose of QoS is to manage contention for network resources to maximize the end-user experience of a session—any kind of session. Because not all packets are equal, they should not be treated equally.
QoS features implement a system of managed unfairness in the network. Some sessions receive priority over other sessions; delay-sensitive sessions bypass queues of packets holding sessions less sensitive to delay; when queuing buffers overflow, packets are dropped on sessions that can recover from the loss or on those that can be eliminated with minimal business impact.
To make space for the packets belonging to high-business impact sessions that cannot tolerate loss without affecting the end-user experience, other sessions are managed (that is, packets are selectively delayed or dropped when contention arises) based on QoS policy decisions implemented in the network.
QoS is near meaningless when implemented on only a segment of the network because the QoE perception is equal to the impairment imposed by the worst-performing segment of the network. QoS is an excellent example of the “only as strong as the weakest link” cliché. QoS should be implemented end to end.
Traffic flowing through your network can be categorized into the following RFC recommended Traffic classes as shown in Figure 1, 2 and 3 below. Figure 4 show how a company or organization can migrate from a 4-class model to an 8-class model then to a 12-class model.
Figure 1: RFC Guidelines for Traffic Classes
Figure 2: Cisco’s RFC 4594-Based Application Class QoS Recommendations
Figure 3 Expanded QoS Model Based on RFC 2597, Clarification
Figure 4: Application-Class Expansion Strategy Example
CAMPUS QoS DESIGN
As far as Campus QoS Design, we begin the exercise applying strategic QoS models to a tactical place in the network (PIN), which in this case is the enterprise campus. Campus-specific design considerations and recommendations are looked at length, taking into account each and every need of our customers. In this section of Campus QoS Design, we specialize in design recommendations for the access, distribution, and core layers of the campus network.
WIRELESS LAN QoS DESIGN
Wireless LAN QoS Design applies the strategic QoS models to the enterprise wireless LAN. Because WiFi is a unique media, as compared to the rest of the network, additional concepts need to be covered and critically looked at to ensure QoS can be achieved over-the-air.
These considerations include the introduction of the Enhanced Distributed Coordination Function as well as IEEE 802.11e/Wireless Multimedia QoS. Our QoS design addresses both the centralized wireless LAN controller deployment model and the new wired-and-wireless converged access deployment model.
The primary role of quality of service (QoS) in the wireless networks is to reduce the latency and jitter of real-time applications over the wireless media. A secondary role of QoS in the WLAN is to provide application-management for traffic originating on wireless devices, including classification (which may incorporate deep packet inspection), marking, and policing. Our job is to ensure that we deliver a Wireless QoS design that works to meet your current and future needs.
DATA CENTER QoS DESIGN
Data Center QoS Design continues the application of QoS strategies, but this time to the data center network. Because of the convergence of storage-area networks and local-area networks within the data center, certain protocols require a completely lossless service that traditional QoS tools cannot guarantee.
Therefore, when it comes to Data Center QoS Design, data center-specific QoS tools are carefully considered and selected, including the data center bridging toolset, which can be leveraged to guarantee such a lossless service. We also address the virtual access layer, access and aggregation layers, and the core layer of data center networks.
WAN AND BRANCH QoS DESIGN
QoS deployments must be implemented end-to-end for it to be successful. Apart from deploying QoS on the Enterprise Campus network, we also design WAN and Branch part of your network as far as QoS design is concerned.
WAN and Branch QoS Design expands the scope of discussion beyond the local area and applies strategic QoS principles to the wide-area network. Our QoS designs are presented for both WAN aggregation routers and for branch routers.
MPLS VPN QoS DESIGN
Our MPLS VPN QoS Design continues the wide-area discussion but addresses
QoS strategies for MPLS VPN networks, taking the perspectives of both the enterprise customer and the service provider into account in the end-to-end design. Our Designs are presented for the enterprise customer-edge router, the provider-edge router and the provider core routers.
IPSEC QoS DESIGN
There is no doubt that Site-to-Site IPsec VPNs are dominating the Enterprises as a form of WAN connection between the remote branches and the main site/HQ of many companies and businesses. And also businesses have opted to use these IPsec VPNs for their VoIP and Video communications and thus the need for QoS over IPsec VPNs.
IPsec QoS Design applies strategic QoS principles to IPsec VPNs. Our QoS designs are detailed for both Dynamic Multipoint VPNs and Group Encrypted Transport VPNs.
PRACTICAL QoS EXAMPLE
QoS NETWORK TOPOLOGY DIAGRAM
Figure 5: ADVANCED QoS NETWORK TOPOLOGY DIAGRAM
QoS is the strongest as its weakest link and that is why a successful QoS deployment is the one that implements QoS end to end. That is to say from one Local Area Network (Enterprise Campus Network) to the other LANs of the same customer over the Service provider network.
In our demonstration we are going to show you the implementation of QoS in the Enterprise Campus Network. This document will be later updated to demonstrate QoS in all of the following areas (in a Service Provider that implements MPLS in its SP backbone.)
1. Enterprise Campus network QoS Implementations
2. Enterprise (CE) to Service Provider (PE) WAN QoS Implementations
3. Service Provider (PE) to Enterprise (CE) WAN QoS Implementations
4. Service Provider Backbone QoS Implementations
ENTERPRISE CAMPUS NETWORK QOS IMPLEMENTATIONS
Here we simply show you of the customer SAFARI SACCO’s Campus network of four switches and how QoS is deployed.
To start us off in this demonstration is the classification of different types of traffic running in this network to different classes. We are going to use 12-class model in putting traffic into different classes.
As a QoS best practice, classification and marking of traffic should happen as close to the source of the traffic as possible. This happens to be in most case the access layer switches, and that is where the intense classification and marking takes place.
Access Control Lists (ACLs) and NBAR/NBAR2 among other methods are used to identify and classify traffic. In this demonstration, we use both.
Since the QoS configuration of the Access Layer switches is identical, there is no need to show the QoS verification on both access layer switches, we will show one of them. The same goes for the collapsed core switches in this network, we will show the QoS configuration for one Core switch.
SAFARI-SACCO-KISII-ACCESS-SW1, Access Layer Switch