6
LISP Interoperability Testing Margaret Wasserman [email protected] IETF 76, Hiroshima, Japan

LISP Interoperability Testing Margaret Wasserman [email protected] IETF 76, Hiroshima, Japan

Embed Size (px)

Citation preview

Page 1: LISP Interoperability Testing Margaret Wasserman mrw@sandstorm.net IETF 76, Hiroshima, Japan

LISP Interoperability Testing

Margaret Wasserman

[email protected]

IETF 76, Hiroshima, Japan

Page 2: LISP Interoperability Testing Margaret Wasserman mrw@sandstorm.net IETF 76, Hiroshima, Japan

Interoperability Tests

• Held in IETF terminal room on Monday evening

• Four implementations represented– Most combinations did interoperate– Two implementation bugs found– Other reasons for lack of interoperability are well-

understood

• Some issues were addressed and interoperability verified later in the week

Page 3: LISP Interoperability Testing Margaret Wasserman mrw@sandstorm.net IETF 76, Hiroshima, Japan

Implementation Status

Cisco LISP

• Cisco NX-OS• Control and data planes• Static and dynamic map

cache entries• Echo nonce, RLOC

probe, RLOC reachability

• IPv4, IPv6, Cross-AFI

OpenLISP

• FreeBSD 7.x kernel• Data plane only• Static map cache

entries• RLOC reachability, map

versioning• IPv4, IPv6, Cross AFI

Page 4: LISP Interoperability Testing Margaret Wasserman mrw@sandstorm.net IETF 76, Hiroshima, Japan

Implementation Status

LISP-Click

• Click Java Framework• Data plane only• Static map cache

entries• Echo nonce, RLOC

reachability• IPv4 only

ZLisp

• Portable C++ (Linux, FreeBSD, MacOS)

• Control and data planes• Dynamic map cache

entries• Responds to RLOC

probes• IPv4, IPv6, Cross AFI

Page 5: LISP Interoperability Testing Margaret Wasserman mrw@sandstorm.net IETF 76, Hiroshima, Japan

Interoperability Matrix

Cisco LISP OpenLISP Yes [1], Data only

Cisco LISP LISP-Click Yes, Data only, IPv4 only

Cisco LISP ZLisp Yes [2], Data & Control

OpenLisp LISP-Click Yes [1], Data only, IPv4 only

OpenLisp ZLisp No [3]

Lisp-Click ZLisp No [3]

[1] After bug fix(es)[2] After RLOC probe handling added to ZLisp[3] Incompatible map cache configuration mechanisms, not yet addressed

Page 6: LISP Interoperability Testing Margaret Wasserman mrw@sandstorm.net IETF 76, Hiroshima, Japan

Possible Specification Issues

• Should it be mandatory to support RLOC Probes?– If not, flag is needed in map reply to

indicate support for RLOC probing

• Support for map request/map reply needs to be mandatory, or static map cache entries need to be mandatory, or both