1 .. This work is licensed under a Creative Commons Attribution 4.0 International License.
2 .. http://creativecommons.org/licenses/by/4.0
4 ==========================================
5 High Priority Traffic Path
6 ==========================================
8 https://wiki.opnfv.org/display/ovsnfv/OVSFV+Requirement+-+High+Priority+Traffic+Path
13 A network design may need to adequately accommodate multiple classes of traffic, each
14 class requiring different levels of service in critical network elements.
16 As a concrete example, a network element managed by a service provider may be
17 handling voice and elastic data traffic. Voice traffic requires that the end-to-end
18 latency and jitter is bounded to some numerical limit (in msec) accuracy in order to ensure
19 sufficient quality-of-service (QoS) for the participants in the voice call.
20 Elastic data traffic does not impose the same demanding requirements on the network
21 (there will be essentially no requirement on jitter. For example, when downloading a
22 large file across the Internet, although the bandwidth requirements may be high there
23 is usually no requirement that the file arrives within a bounded time interval.
25 Depending on the scheduling algorithms running on the network element,
26 frames belonging to the data traffic may get transmitted before frames
27 belonging to the voice traffic introducing unwanted latency or jitter.
28 Therefore, in order to ensure deterministic latency and jitter characteristics
29 end-to-end, each network element through which the voice traffic traverses
30 must ensure that voice traffic is handled deterministically.
32 Hardware switches have typically been designed to ensure certain classes
33 of traffic can be scheduled ahead of other classes and are also
34 over-provisioned which further ensures deterministic behavior when
35 handling high priority traffic. However, software switches (which includes
36 virtual switches such as Open vSwitch) may require modification in order
37 to achieve this deterministic behavior.
42 1. Program classes of service
44 The End User specifies a number of classes of service. Each class of service
45 will be represented by the value of a particular field in a frame. The class
46 of service determines the priority treatment which flows in the class will
47 receive, while maintaining a relative level of priority for other classes and
48 a default level of treatment for the lowest priority class of service. As
49 such, each class of service will be associated with a priority. The End User
50 will associate classes of service and priorities to ingress ports with the
51 expectation that frames that arrive on these ingress ports will get
52 scheduled following the specified priorities.
54 Note: Priority treatment of the classes of service cannot cause any one of
55 the classes (even the default class) from being transferred at all. In other
56 words, a strict priority treatment would likely not be successful for serving
57 all classes eventually, and this is a key consideration.
59 2. Forward high priority network traffic
61 A remote network element sends traffic to Open vSwitch. The remote network
62 element, indicates the class of service to which this flow of traffic belongs
63 to by modifying a pre-determined but arbitrary field in the frame as specified
64 in Use Case 1. Some examples include the Differentiated Services Code Point
65 (DSCP) in an IP packet or the Priority Code Point (PCP) in an Ethernet frame.
66 The relative priority treatment that frames get processed by Open vSwitch can be guaranteed by the
67 values populated in these fields when the fields are different. If the fields
68 are the same, ordering is not deterministic.
70 For example: Packet A is sent with a DSCP value of 0 and packet B is sent
71 with a value of 46; 0 has a lower priority than 46. Packet A arrives
72 before packet B. If Open vSwitch has been configured as such, Packet
73 B will be transmitted before Packet A.
100 Other end user impact
101 ---------------------
110 Other deployer impact
111 ---------------------
126 Who is leading the writing of the code? Or is this a blueprint where you're
127 throwing it out there to see who picks it up?
129 If more than one person is working on the implementation, please designate the
130 primary author and contact.
151 In order to test how effectively the virtual switch handles high priority traffic
152 types, the following scheme is suggested.::
154 +---------------------------+ Ingress Traffic Parameters
155 | | +-------------------------------------------+
157 | | Packet Size: The size of the Ethernet frames
159 | | Tmax: RFC2544 Max. Throughput for traffic of
160 | PHY0 <-------+ "Packet Size"
162 | | Total Offered Rate: The offered rate of both
163 | | traffic classes combined expressed as a % of
166 | | Ingress Rates are expressed as a percentage
167 | | of Total Offered Rate.
170 | OVS | Ethernet PCP = 0 (Background)
171 | (BR0) | Ingress Rate : rate_ingress_a(n) Mfps
174 | | Ethernet PCP = 7 (Highest)
175 | | Ingress Rate : rate_ingress_b(n) Mfps
177 | | Egress Traffic Measurements
178 | | +-------------------------------------------+
180 | | Egress Throughput : rate_egress_a(n) Mfps
181 | | Egress Latency : max_lat_egrees_a(n) ms
182 | | Egress Jitter : max_jit_egress_a(n) ms
185 | | Egress Throughput : rate_egress_b(n) Mfps
186 | | Egress Latency : max_lat_egrees_b(n) ms
187 +---------------------------+ Egress Jitter : max_jit_egress_b(n) ms
190 Open vSwitch is configured to forward traffic between two ports agnostic to the
191 traffic type. For example, using the following command:
193 ovs-ofctl add-flow br0 in_port=0,actions=output:1
195 The test will be carried out with the functionality to enable high-priority
196 traffic enabled and disabled in order to guage the change in performance for
199 Two classes of traffic will be generated by a traffic generator. In the example
200 above, the classes are differentiated using the Ethernet PCP field. However,
201 another means for differentiating traffic could be used, depending the
202 prioritization scheme that is developed.
204 Tests should be performed for each combination of:
206 * Packet Sizes in (64, 512)
207 * Total Offered Rate in (80, 120, 150)
208 * rate_ingress_b(n) / rate_ingress_a(n) in (0.1, 0.2, 0.5)
210 For each set, the following metrics should be collected for each traffic
211 class over a specified time period:
213 Egress Throughput (Mfps)
214 Maximum Egress Latency (ms)
215 Maximum Egress Jitter (ms)
225 Please add any useful references here. You are not required to have any
226 reference. Moreover, this specification should still make sense when your
227 references are unavailable. Examples of what you could include are:
229 * Links to mailing list or IRC discussions
231 - http://lists.opnfv.org/pipermail/opnfv-tech-discuss/2015-December/007193.html
232 - http://ircbot.wl.linuxfoundation.org/meetings/opnfv-ovsnfv/2016/opnfv-ovsnfv.2016-03-07-13.01.html
234 * Links to relevant research, if appropriate
236 - https://wiki.opnfv.org/download/attachments/5046510/qos_mechanisms.pdf?version=1&modificationDate=1459187636000&api=v2
238 * Related specifications as appropriate
240 * Anything else you feel it is worthwhile to refer to
246 Optional section intended to be used each time the spec
247 is updated to describe new design, API or any database schema
248 updated. Useful to let reader understand what's happened along the
251 .. list-table:: Revisions