Page History

Table of contents

...

Note that in modern network devices, the network interfaces are usually combined and can operate both as incoming and outgoing.

...

With QoS configured, each of the incoming traffic flows can be classified by its type (for example) and a separate queue can be mapped to each class (Figure 2b). Each packet queue can be assigned a priority, which will be taken into account while extracting the packets from the queues, and will guarantee specific quality indicators. The traffic flow classification can be performed not only with respect to the services used, but according to other criteria also. For example, each pair of nodes can be assigned to a separate packet queue (Figure 2c).

...

• The throughput of the communication channel and of the network devices is limited.
• The data delivery time from source to destination is non-zero.
• A communication channel is a medium with a set of physical parameters that determine can have influence on the signal propagation effects.
• The software and hardware architecture of the network devices impacts the way in which the data is being transmitted.

...

• Losses in the medium: losses related with the propagation of the signal in the physical environment. For example, the frame will be lost if the useful signal level is lower than the receiver sensitivity. Losses can also be caused by the physical damage of the interfaces connected to the media or by impulse pickups resulting from poor grounding.
• Losses on the interface: losses while processing a queue at the incoming or at the outgoing interface. Each interface has a memory buffer, which can be completely filled in case of intensive data stream transmissions. In this case, all the subsequent data entering the interface will be discarded, because it cannot be buffered.
• Losses inside the device: Data discarded by the network device according to the logic of the configuration. If the queues are full and the incoming data cannot be added to the processing queue, the network device will drop it. Also, these losses include the data packets rejected by access lists and by the firewall.

...

For example, the service header length for 64 bytes long frames (Figure 4b) and 156 bytes frames(Figure 4c) will be the same, but the user data amount will be different. To transmit 138 bytes of user data, three 64 bytes frames or one 156 bytes frame will be required, so in the first case 192 bytes are sent, in the second only 156 bytes. For a link having a fixed throughput, large frames will increase the efficiency by rising the useful throughput of the system, but the latency will also increase. The performance of the Infinet devices in various conditions is shown in the "Performance of the InfiNet Infinet Wireless devices" document.

...

The delay is often measured, as a round-trip time (RTT), i.e. the time it takes for the data packet to be transmitted from the source to the destination and backward. For example, this value can be seen in the ping command's results. The time it takes for the intermediate network devices to process the data packets forward and backward may differ, therefore, usually the round-trip time is not equal to the double of the one-way delay.

Jitter

The CPU load and the status of the packet queues status are frequently changing at the intermediate network devices, so the delay during the data packet transmission will vary. In the example below (Figure 6), the transmission time for the packages packets with the identifiers 1 and 2 is different. The difference between the maximum and the average delay values is called jitter.

...

The effect depends on the characteristics of the service and on the ability of the higher layer network protocols to restore the original sequence. For example Usually, if the traffic of different services is transmitted through different paths, then it should not affect the ordering of the received data.

...

Each of the data transfer services has a set of requirements for the quality indicators. The RFC 4594 document includes the following service types:

QoS methods

The traffic transmission of the various services is performed on a single network infrastructure, which has limited resources, therefore, mechanisms should be provided for distributing the resources between the services.

Let's look at the example below (Figure 8). Node-2 generates traffic serving different services with a total speed of 1 Gbit/s. Medium-2 allows to transfer this data stream to an intermediate network device, however, the maximum link throughput between the Network device and Node-5 is 500 Mbps. Obviously, the data stream cannot be processed completely and part of this stream must be dropped. The QoS task is to make these drops manageable in order to provide the required metric values for the end services. Of course, it is impossible to provide the required performance for all the services, as the throughput does not match, therefore, the QoS policy implementation involves that the traffic of the the critical services should be processed first.

The example above allows to highlight two Two main methods used during the QoS policy implementation can be highlighted:

• Prioritization: data distribution among  the distribution of the data packets into queues and the extraction of the packets selection from the queues by their priority. In this case, the packets that are most sensitive to delay and jitter are processed first, then the traffic for which the delay value is not critical is processed.
• Throughput limitation: throughput limitation for the traffic flows. All the traffic that exceeds the set throughput threshold will be discarded.

...

• Step 1:
  • Node-1 and Node-2 generate packets for two services: telephony and mail. The telephony traffic is sensitive to delay and jitter unlike the mail service data (see Services requirements for quality indicators), therefore, it must be processed first by the intermediate devices.
  • Network device-1 receives the packets of Node-1 and of Node-2.
• Step 2:
  • Traffic prioritization is configured on Network device-1, thus the device classifies the incoming traffic and places the data packets in different queues. All the voice traffic will be put in queue 0, and the mail traffic will be put in queue 16. Thus, the priority of queue 0 is higher than the one of queue 16.
  • The packets leave the queues and proceed towards the outgoing interfaces in accordance with the queue priorities i.e. queue 0 will be emptied first, then queue 16 will be emptied.
• Step 3:
  • Network device-1 sends data to Medium-7, which is connected with the Network device-2. The sequence of data packets corresponds to the quality metrics - the telephony data is transmitted first through the medium, and the mail service is sent next.
  • Node-3 is connected to Network device-2 and generates a mail service data stream.
• Step 4:
  • Network Device-2 has no prioritization settings, thus all the incoming traffic is put in queue 16. The data will leave the queues in the same order that it entered, i.e. the telephony and the mail services will be handled equally, despite the requirements of the quality indicators.
  • Network device-2 increases the delay for the telephony traffic transmission.
• Step 5:
  • The data is transmitted to the end nodes. The transmission time of the voice packets was also increased due to the additional processing of the mail service traffic of Node-3.

...

Keep in mind that implementing QoS policies is the only method to ensure the quality metrics. For a maximum an optimal effect, the QoS configuration should be synchronized with other settings. For example, using the TDMA technology instead of Polling on the InfiLINK 2x2 and InfiMAN 2x2 families of devices reduces jitter by stabilizing the value of the delay value  (see TDMA and Polling: Application features).

The traffic prioritization mechanism

...

• White-box: all the network devices along the data propagation path are in the same responsibility zone. In this case, the QoS configuration on the devices can be synchronized, according to the requirements specified in the section above.
• Black-box: some network devices in the data propagation path are part of an external responsibility zone. The classification rules for incoming data and the algorithm for emptying the queues are configured individually on each device. The architecture of the packet queues's implementation depends on the manufacturer of the equipment, therefore there is no guarantee of a correct QoS configuration on the devices in the external responsibility zone, and as a result, there is no guarantee of the high-quality performance indicators.

...

Keep in mind that usually, the equipment located in an external responsibility zone does not support data prioritization in accordance with the priority values in the service headers. Traffic priority coordination should be performed at the border of the responsibility zones and , at the administrative level, by implementing additional network configuration settings.

...

The Ethernet frame header includes the "User Priority" service field, which is used to prioritize the data frames. The field has a size of 3 bits size, which allows to select 8 traffic classes: 0 - the lowest priority class, 7 - the highest priority class. Keep in mind that the "User Priority" field is present only in 802.1q frames, i.e. frames using VLAN tagging.

...

1. When the protocol was first approved, there was an 8-bit ToS (Type of Service) field in the IP packet header (see RFC 791). ToS included the following fields (Figure 12a):
   1. Precedence: priority value (3 bits).
   2. Delay: delay minimization bit.
   3. Throughput: throughput minimization bit.
   4. Reliability: reliability maximization bit.
   5. 2 bits equal to 0.
2. In the second stage, 8 bits were still used for packets packet prioritization, however, ToS included the following fields (see RFC 1349):
   1. Delay.
   2. Throughput.
   3. Reliability.
   4. Cost: bit to minimize the cost metric (1 bit is used, whose value was previously zero).
3. Last, the IP header structure has been changed (see RFC 2474).The 8 bits previously used for prioritization were distributed in the following way (Figure 12b):
   1. DSCP (Differentiated Services Code Point): packet priority (6 bits).
   2. 2 bits are reserved.

Thus, ToS allows to distinguish 8 traffic classes: 0 - the lowest priority, 7 - the highest priority, and DSCP - 64 classes: 0 - the lowest priority, 63 - the highest priority.

...

Let's look at the example of a data transmission from Node-1 to Node-2 through a DS-domain and through a third-party telecom operator's network (Figures 13a-c). The DS domain includes three devices, two of them are located at the borderline and one is an intermediate device. Lets  Let's look at the steps of taken for data processing in a network using an Ethernet frame transmission (the basic principles discussed in the example below are applicable for an IP packet or other protocol that supports data prioritization):

• Step 1: Node-1 generates an Ethernet frame for Node-2. There is no field present for frame priority tagging in the header (Figure 13a).
• Step 2: Border Network Device-1 changes the Ethernet header, setting the priority to 1. Border devices should have rules configured in order to filter the traffic of Node-1 from the general stream and assigna  to assign a  priority for it (usually the traffic of several devices is aggregated using a switch before reaching Border network device-1, but this is a simplified example). In networks with a large traffic flow number, the list of rules on border devices can be volumetric. Border network device-1 processes the frame according to the set priority, placing it in the corresponding queue. The frame is transmitted towards the outgoing interface and sent to Intermediate network device-2 (Figure 13a).
• Step 3: Intermediate network device-2 receives the Ethernet frame having priority 1, and places it in the corresponding priority queue. The device does not handle the priority in terms of changing or removing it inside the frame header. The frame is next transmitted towards the Border network device-3 (Figure 13a).
• Step 4: Border network device-3 processes the incoming frame similarly to the Intermediate device-2 (see Step 3) and forwards it towards the service network provider(Figure 13a).
  • Step 4a: in case of agreeing that the traffic will be transmitted through the provider's network with a priority other than 1, then Border Device-3 must perform a change the priority change. In this example, the device changes the priority value from 1 to 6 (Figure 13b).
• Step 5: during the transmission of the frame through the provider's network, the devices will take into account the priority value in the Ethernet header (Figure 13a).
  • Step 5a: similarly to Step 4a (Figure 13b).
  • Step 5b: if there is no agreement about the frame prioritization according to the priority value specified in the Ethernet header, a third-party service provider can apply its own QoS policy and set a priority that may not satisfy the QoS policy of the DS domain (Figure 13c).
• Step 6: the border device in the provider's network removes the priority field from the Ethernet header and forwards it to Node-2 (Figure 13a-c).

...

• The devices automatically recognize priorities according to different protocols. For example, the InfiLINK XG family of devices supports 802.1p prioritization, but does not recognize DSCP priority values.
• The devices at the borderline of the DS domain allow to use a different set of criteria to classify the traffic. For example, the InfiMAN 2x2 devices allow to set priorities by selecting all the TCP traffic directed to port 23, while the Quanta 5 family devices does not support this type of prioritization.
• The number of the queues implemented inside the devices differs and depends on the manufacturer. A correspondence table is used to set a relation between the priority in the service header and the device's internal queue.

...

Please note the architectural queuing feature of the Infinet devices: all queues share a single memory buffer. In case that all the traffic falls into a single queue, its the size of the queue will be equal to the size of the buffer, but if there will be several queues in use, the size of the memory buffer will be evenly divided between them.

...

Prioritization assumes the use of several packet queues, whose content must be transmitted to the outgoing interfaces through a common bus. Infinet devices support two mechanisms for packets packet transmission from the queues to the bus: strict and weighted scheduling.

...

The disadvantage of this mechanism is that resources will not be allocated to low-priority traffic if there are packets in the higher priority queues, leading to the complete inaccessibility of some network services.

...

The weighted scheduling doesn't have the disadvantages of the strict scheduling. Weighted scheduling assumes the allocation of the resources between for all the queues according to the weighting factors that correspond to the priority values. If there are three queues (Figure 15), weighted factors can be distributed in the following way:

• packets packet queue 1: weight = 3;
• packets packet queue 2: weight = 2;
• packets packet queue 3: weight = 1.

When using the weighted scheduling, each queue will receive resources, i.e. there will be no such situation with complete inaccessibility of a network service.

...

• Pay special attention when developing the QoS policies. The policy should take into account the traffic of all the services used in the network and it should provide strict compliance between the service and the traffic class.
• The QoS policy should take into account the technical capabilities of the devices for recognizing and manipulating handling the service field values, which indicate the data priority.
• The rules for traffic flow classification must be configured on the border devices of the DS domain.
• The  intermediate intermediate devices of the DS domain should automatically recognize the traffic priorities.

Throughput limitation mechanism 

The distribution of the network resource distribution resources between the traffic flows can be performed not only by prioritization, but also using the throughput limitation mechanism. In this case, the bitrate of the stream cannot exceed the threshold level set by the network administrator.

...

The throughput limitation principle is to constantly measure the throughput of the data stream intensity and to apply restrictions if the intensity this value exceeds the set threshold (Figure 16a,b). The throughput limitation in Infinet devices is performed according to the Token Bucket algorithm, where all data packets above the throughput threshold are discarded. As a result, there will appear losses, as described above.

Token Bucket Algorithm

For each speed limit rule there is a logical buffer associated, in order to serve the allowed amount of transmitted data. Usually, the buffer size is larger than the size of the limitation. Each unit of time is allocated to a data size equal to the threshold for the bitrate limit.

In the example below (video 1), the speed limit is 3 data units and the buffer size is 12 data units. The buffer is constantly filled in according to the threshold, however, it cannot be filled over its own size.

...

The data received by the device at the inbound interface will be processed only if the buffer contains has resources for their processing (video 2). Thus, the passing data occupies the buffer's resources. If the buffer's resources are fully used occupied at the time of a new data frame arrival, the frame will be discarded.

...

The rate of the data flows in packet networks is inconsistent, proving the efficiency of the Token Bucket algorithm.  Time The time intervals in which data is not transmitted allows to accumulate resources in the buffer, and then process the amount of data that exceeds the threshold. A wide band will be allocated to pulse data streams, such as web traffic, in order to ensure a quick loading of the web pages and to increase the comfort level of the end user.

...

The Token Bucket algorithm can be applied to separate traffic flows. In this case, a speed limit buffer will be allocated for each flow (video 4).

In this example, two speed limit rules are implemented: for the traffic of vlan 161 - 3 data units per unit of time block, for the traffic of vlan 162 - 2 data units. The buffer size for each traffic flow contains 4 time intervals, i.e. 12 data units for vlan's 161 traffic and 8 data units for vlan's 162 traffic. In total, 5 data units are allocated to the buffers in each time interval, then the allocated resources are distributed between the buffers. Since the size of the buffers is limited, the resources that exceed their size cannot be used.

...

There are ways to combine the resource buffers. For example, on the Infinet devices, the allocated resource buffers can be combined using classes (see below). If one resource buffer is filled with resources (video 6), its further incoming resources can be provided to another buffer.

In the example below, the buffer for vlan 162 is full of resources, allowing to fill in the vlan's 161 traffic buffer with 5 data units of resources, instead of 3 (its own 3 data units plus the 2 data units of the other buffer). In this case, the vlan's 161 service throughput will increase. But when vlan's 162 traffic resource buffer will have free space, the resource allocation will return to the normal mode: for vlan's 161 traffic buffer - 3 data units, for vlan's 162 traffic buffer - 2 data units.

Anchor
shaping_infinet shaping_infinet

Types of speed limits in Infinet devices

...

• Traffic shaping at the physical interface: limitations will be applied to the whole data flow passing through the physical interface. This method is easy to configure - specify the interface and the threshold value , - but it does not allow to apply limitations to a specific network service traffic.
• Traffic flow shaping: limitations are applied to the logical data flows. The logical data stream is separated from the main traffic by a specified criteria. It allows to apply throughput limitations per network services, which are separated by the values of the service header fields. For example, the traffic tagged with vlan 42 can be separated to a logical channel and limited in throughput without influencing the other traffic flows.

...

Let's look at the example of transmitting the traffic of two services associated with vlan id's 161 and 162, between Master and Slave (Figure 17a).   The total traffic of the services should not exceed 9 Mbps.

...

• Class 16 has been configured with a 9 Mbps throughput.
• Class 16 is the parent of the channels 161 and 162, i.e. the total traffic at these logical channels is limited to 9 Mbps.
• The traffic with vlan ID 16 is associated with the logical channel 161; the traffic of vlan 162 is associated with the logical channel 162.
• The CIR value for channel 161 is 4 Mbps and for channel 162 it is 5 Mbps. If both services will actively exchange data, the threshold values for their traffic will be equal to the CIR of each channel.
• The MIR value for channel 161 is 9 Mbps and for channel 162 it is 7 Mbps. If there is no traffic in the logical channel 162, then the threshold value for channel 161 will be equal to the MIR, i.e. 9 Mbps. In the other case, when there is no traffic in the logical channel 161, the threshold value for channel 162 will be equal to 7 Mbps.

The throughput limitation capabilities of all Infinet families of devices are shown in the table below:

Recommendations for the throughput limitation configuration 

...

• The traffic of all network services should be limited. It allows to take control over all traffic flows and separately allocate resources for these flows.
  
• The throughput limitation should be performed on the devices closest to the data source. There is no need to duplicate throughput limiting rules for the data flows throughout the chain of intermediate devices.
• Many network services are bidirectional, so they require restrictions on devices for both the incoming and the outgoing traffic.
• To set the correct throughput threshold values, evaluate first the average and the maximum values of the service traffic. Pay special attention to the most busy hours. Collecting data for analysis is possible via the InfiMONITOR monitoring system.
• The sum of the CIR values of the logical channels associated with one class should not exceed the maximum class throughput.

...

1. RFC 4594.
2. RFC 791.
3. RFC 1349.
4. RFC 2474.
5. InfiMONITOR monitoring system.
6. InfiLINK 2x2, InfiMAN 2x2 family devices web interface. QoS options.
7. InfiLINK 2x2, InfiMAN 2x2 family devices web interface. Traffic shaping.
8. InfiLINK Evolution, InfiMAN Evolution family devices web interface. QoS options.
9. InfiLINK Evolution, InfiMAN Evolution family devices web interface. Traffic shaping.
10. InfiLINK XG, InfiLINK XG 1000 family devices web interface. Configuring QoS.
11. Quanta 5, Quanta 6 family devices web interface. Switch settings.
12. Quanta 70 family devices web interface. Switch settings.
13. QoS configuration in OS WANFleX.