Motorola 6806800C47B 用户手册
Checkpoint Service Programmer’s Reference (6806800C47B)
Introduction
Models and Concepts
14
1.2
Models and Concepts
The Checkpoint service comprises three distributed subparts that maintain the cluster-wide
checkpoint database.
checkpoint database.
z
Checkpoint Director
z
Checkpoint Node Director
z
Checkpoint Agent
1.2.1
Checkpoint Director
Checkpoint Director (CPD) runs as a process on a system manager node. CPD maintains the
centralized repository of control information for all checkpoints created in the cluster. The CPD
also maintains the location information of active replicas for all the checkpoints opened in the
cluster. In case of non-collocated checkpoint, the CPD designates a particular node to manage
an active replica for that checkpoint and also decides on the number or replicas to be created
which depends on the policy (See section 8.1.5.1 Usage of Non-Collocated Checkpoints, for
policies). Two instances of CPD are configured, one on each system manager node, in order to
achieve high-availability. The two instances are configured to be part of a service group having
a 2N redundancy model.
centralized repository of control information for all checkpoints created in the cluster. The CPD
also maintains the location information of active replicas for all the checkpoints opened in the
cluster. In case of non-collocated checkpoint, the CPD designates a particular node to manage
an active replica for that checkpoint and also decides on the number or replicas to be created
which depends on the policy (See section 8.1.5.1 Usage of Non-Collocated Checkpoints, for
policies). Two instances of CPD are configured, one on each system manager node, in order to
achieve high-availability. The two instances are configured to be part of a service group having
a 2N redundancy model.
1.2.2
Checkpoint Node Director
There is one instance of the Checkpoint Node Director (CPND) on each system manager and
payload nodes. It is modeled as a separate process. CPND maintains the detailed information
of the Checkpoints referred from that node and the corresponding updates and retrievals that
operate on those checkpoints. CPND also handles the requests issued by the CPA instances
on behalf of its client applications on the same node. In case of checkpoints that have been
created with the collocated attribute and the asynchronous update option, the application will
payload nodes. It is modeled as a separate process. CPND maintains the detailed information
of the Checkpoints referred from that node and the corresponding updates and retrievals that
operate on those checkpoints. CPND also handles the requests issued by the CPA instances
on behalf of its client applications on the same node. In case of checkpoints that have been
created with the collocated attribute and the asynchronous update option, the application will
Figure 1-1
Checkpoint Service -Subparts
C P D
active
C P N D
C P A
A pplication
C P D
standby
P aylo ad B lad e 1
N o d e S M H 1
C P N D
C P A
C P N D
C P A
C P N D
C P A
N o d e S M H 2
A pplication
A pplication
A pplication
P aylo ad B lad e 2
C P D - C h eckp o in t D irecto r
C P A - C h eckp o in t A g en t
S M H - S ystem M an ag er H o st
C o m m u n icatio n p ath (M D S )
C o m m u n icatio n p ath (M B C S v)
C P N D - C h eckp o in t N o d e D irecto r