If you want your switch to support full functionality, you have to consider all
the fileds of the header. Otherwise and for experiments, you can support only a
subset of the header fields. You have to be careful in raising an error, or
omitting extra field when controller request adding or modifying flow entries
that contain header field which your switch doesn't support.
In openflow switch v1.4, tables have the same structure, but may contain
different flow entries based on the functionality.
On Thursday, 6 November 2014, 15:36, me a <mehdi_am...@hotmail.com> wrote:
<!--#yiv7275046566 .yiv7275046566hmmessage
P{margin:0px;padding:0px;}#yiv7275046566
body.yiv7275046566hmmessage{font-size:12pt;font-family:Calibri;}-->
<!--#yiv7275046566 .yiv7275046566ExternalClass .yiv7275046566ecxhmmessage P
{padding:0px;}#yiv7275046566 .yiv7275046566ExternalClass
body.yiv7275046566ecxhmmessage {font-size:12pt;font-family:Calibri;}-->I have
some questions about OpenFlow protocol.If we want to design and implement an
OpenFlow switch according toversion 1.0, header fields of flow table must be
exactly 12 tuples??? Infact we cannot have an OpenFlow switch with 5 tuples? In
other words, headerfields of flow tables are considered by OpenFlow
protocol?Another question is that, in version 1.1 to 1.4, we can have morethan
one table in our design; the question is that each table should have allthe
fields? for example if we have 5 tables and the number of header fieldsthat are
recommended by protocol is 18 tuples,Each table in this design should have 18
tuples or they aresegregated by their functions?Best Regards,Amini.
_______________________________________________
openflow-discuss mailing list
openflow-discuss@lists.stanford.edu
https://mailman.stanford.edu/mailman/listinfo/openflow-discuss
_______________________________________________
openflow-discuss mailing list
openflow-discuss@lists.stanford.edu
https://mailman.stanford.edu/mailman/listinfo/openflow-discuss
