]> git.proxmox.com Git - mirror_ovs.git/blob - OPENFLOW-1.1+
tunnel: Use ofport_dpif instead of ofport.
[mirror_ovs.git] / OPENFLOW-1.1+
1 OpenFlow 1.1+ support in Open vSwitch
2 =====================================
3
4 Open vSwitch support for OpenFlow 1.1, 1.2, and 1.3 is a work in
5 progress. This file describes the work still to be done.
6
7 The Plan
8 --------
9
10 OpenFlow version support is not a build-time option. A single build
11 of Open vSwitch must be able to handle all supported versions of
12 OpenFlow. Ideally, even at runtime it should be able to support all
13 protocol versions at the same time on different OpenFlow bridges (and
14 perhaps even on the same bridge).
15
16 At the same time, it would be a shame to litter the core of the OVS
17 code with lots of ugly code concerned with the details of various
18 OpenFlow protocol versions.
19
20 The primary approach to compatibility is to abstract most of the
21 details of the differences from the core code, by adding a protocol
22 layer that translates between OF1.x and a slightly higher-level
23 abstract representation. The core of this approach is the many struct
24 ofputil_* structures in lib/ofp-util.h.
25
26 As a consequence of this approach, OVS cannot use OpenFlow protocol
27 definitions that closely resemble those in the OpenFlow specification,
28 because openflow.h in different versions of the OpenFlow specification
29 defines the same identifier with different values. Instead,
30 openflow-common.h contains definitions that are common to all the
31 specifications and separate protocol version-specific headers contain
32 protocol-specific definitions renamed so as not to conflict,
33 e.g. OFPAT10_ENQUEUE and OFPAT11_ENQUEUE for the OpenFlow 1.0 and 1.1
34 values for OFPAT_ENQUEUE. Generally, in cases of conflict, the
35 protocol layer will define a more abstract OFPUTIL_* or struct
36 ofputil_*.
37
38 Here are the current approaches in a few tricky areas:
39
40 * Port numbering. OpenFlow 1.0 has 16-bit port numbers and later
41 OpenFlow versions have 32-bit port numbers. For now, OVS
42 support for later protocol versions requires all port numbers to
43 fall into the 16-bit range, translating the reserved OFPP_* port
44 numbers.
45
46 * Actions. OpenFlow 1.0 and later versions have very different
47 ideas of actions. OVS reconciles by translating all the
48 versions' actions (and instructions) to and from a common
49 internal representation.
50
51 OpenFlow 1.1
52 ------------
53
54 The list of remaining work items for OpenFlow 1.1 is below. It is
55 probably incomplete.
56
57 * Implement Write-Actions instruction.
58
59 * The new in_phy_port field in OFPT_PACKET_IN needs some kind of
60 implementation. It has a sensible interpretation for tunnels
61 but in general the physical port is not in the datapath for OVS
62 so the value is not necessarily meaningful. We might have to
63 just fix it as the same as in_port.
64
65 * On OF1.1+ flow_mods, updates by MODIFY are now much better
66 specified. Check that OVS implements the new behavior, fix it
67 if not.
68
69 * OFPT_TABLE_MOD stats. This is new in OF1.1, so we need to
70 implement it. It should be implemented so that the default OVS
71 behavior does not change.
72
73 * MPLS. Simon Horman maintains a patch series that adds this
74 feature. This is partially merged.
75
76 * SCTP. Joe Stringer maintains a patch series that adds this
77 feature. It has received review comments that need to be
78 addressed before it is merged.
79
80 * Match and set double-tagged VLANs (QinQ). This requires kernel
81 work for reasonable performance.
82
83 * VLANs tagged with 88a8 Ethertype. This requires kernel work for
84 reasonable performance.
85
86 * Groups.
87
88 OpenFlow 1.2
89 ------------
90
91 OpenFlow 1.2 support requires OpenFlow 1.1 as a prerequisite, plus the
92 following additional work. (This is based on the change log at the
93 end of the OF1.2 spec. I didn't compare the specs carefully yet.)
94
95 * OFPT_FLOW_MOD:
96
97 * MODIFY and MODIFY_STRICT commands now never insert new flows
98 in the table. We will still need variations that do,
99 though, both to support older OpenFlow protocols and to get
100 sensible behavior for the internal implementation of the
101 NXAST_LEARN action.
102
103 * New flag OFPFF_RESET_COUNTS.
104
105 * New cookie field behavior.
106
107 * Add ability to delete flow in all tables.
108
109 * Update DESIGN to describe OF1.2 behavior also.
110
111 OpenFlow 1.3
112 ------------
113
114 OpenFlow 1.3 support requires OpenFlow 1.2 as a prerequisite, plus the
115 following additional work. (This is based on the change log at the
116 end of the OF1.3 spec, reusing most of the section titles directly. I
117 didn't compare the specs carefully yet.)
118
119 * Add support for multipart requests.
120
121 * Add OFPMP_TABLE_FEATURES statistics.
122
123 * More flexible table miss support.
124
125 * IPv6 extension header handling support. Fully implementing this
126 requires kernel support. This likely will take some careful and
127 probably time-consuming design work. The actual coding, once
128 that is all done, is probably 2 or 3 days work.
129
130 * Per-flow meters. Similar to IPv6 extension headers in kernel
131 and design requirements. Might be politically difficult to add
132 directly to the kernel module, since its functionality overlaps
133 with tc. Ideally, therefore, we could implement these somehow
134 with tc, but I haven't investigated whether that makes sense.
135
136 * Per-connection event filtering. OF1.3 adopted Open vSwitch's
137 existing design for this feature so implementation should be
138 easy.
139
140 * Auxiliary connections. These are optional, so a minimal
141 implementation would not need them. An implementation in
142 generic code might be a week's worth of work. The value of an
143 implementation in generic code is questionable, though, since
144 much of the benefit of axuiliary connections is supposed to be
145 to take advantage of hardware support. (We could make the
146 kernel module somehow send packets across the auxiliary
147 connections directly, for some kind of "hardware" support, if we
148 judged it useful enough.)
149
150 * MPLS BoS matching. (Included in Simon's MPLS series?)
151
152 * Provider Backbone Bridge tagging. I don't plan to implement
153 this (but we'd accept an implementation).
154
155 * Rework tag order. I'm not sure whether we need to do anything
156 for this.
157
158 * Duration for queue stats. (Duration for port stats is already
159 implemented.)
160
161 * On-demand flow counters. I think this might be a real
162 optimization in some cases for the software switch.
163
164 How to contribute
165 -----------------
166
167 If you plan to contribute code for a feature, please let everyone know
168 on ovs-dev before you start work. This will help avoid duplicating
169 work.
170
171 Please consider the following:
172
173 * Testing. Please test your code.
174
175 * Unit tests. Please consider writing some. The tests directory
176 has many examples that you can use as a starting point.
177
178 * ovs-ofctl. If you add a feature that is useful for some
179 ovs-ofctl command then you should add support for it there.
180
181 * Documentation. If you add a user-visible feature, then you
182 should document it in the appropriate manpage and mention it in
183 NEWS as well.
184
185 * Coding style (see the CodingStyle file at the top of the source
186 tree).
187
188 * The patch submission guidelines (see SubmittingPatches). I
189 recommend using "git send-email", which automatically follows a
190 lot of those guidelines.
191
192 Bug Reporting
193 -------------
194
195 Please report problems to bugs@openvswitch.org.