diffrence between multicast and unicast
- Life Logs/Study Notes
- 2014. 7. 30. 07:00
안녕하세요, 우리노트 Benjamin입니다.
HACMP 7.1 부터는 node간 heartbeat check 를 위하여 Multicast 를 사용합니다.
하지만 multicast 가 오작동하면서 생기는 문제들이 많이 발견되어서.. 7.1.3 에서 다시 unicast로 변경되죠..
더 큰 문제는 multicast 와 unicast 의 차이를 잘 모르겠다는겁니다.
그래서 찾아보니 아래와 같은 문서 내용이 있더군요.
Problem(Abstract)
You want to know the differences between unicast and multicast.
Resolving the problem
IBM supports a "point to point" publish and subscribe model. This can be one to many, many to one, or many to many. This is sometimes referred to as a "unicast" model.
Broadcast is a protocol that allows a sender or publisher to issue a message once which gets propagated to all end points as defined by the underlying LAN infrastructure and router configurations.
Multicast is another protocol (supported with WMQ V7.1 or later) which propagates a message to a sub-group of end points – again defined by the underlying LAN infrastructure and router configuration. In multicast and other 'spray' technologies the end points that do not want the information have to be shielded or insulated from receiving these unwanted messages.
In essence the use of multicast is limited to niche markets such as trading floors. A "unicast" model is far more flexible because it allows subscribers to change, as often as they like, the information that is of interest to them.
요약해보면..
unicast 는 point to point 로 통신하는 방식이라면,
Multicast는 새로운 통신 규격으로서 자신에게 등록된 하위그룹에게 동시에 통신하는 Spray방식입니다.
그리고 unicast가 more flexible 하다는 내용도 적혀 있군요.. 되도록 PowerHA 7.1.3.을 사용하는 편이 좋겠네요..
추가적으로 multicast 에서 ping test 하는 방법 첨부하여 드립니다.
# ping test 할 node (target)들을 receiver 로 지정
HAtest1@/> mping -r
mping version 1.1
Listening on 227.1.1.1/4098:
# ping test 할 node (source)를 sender 로 지정
HAtest2@/> mping -s
mping version 1.1
mpinging 227.1.1.1/4098 with ttl=1:
32 bytes from 172.16.10.230 seqno=0 ttl=1 time=0.310 ms
32 bytes from 172.16.10.230 seqno=1 ttl=1 time=0.312 ms
sender 쪽에서 mping 을 때리면 그때부터 receiver쪽도 pinging 결과가 출력됩니다.