Performing partial subnet initialization in a middleware machine environment

Patent metadata

Field	Value
Publication number	US-9906429-B2
Application number	US-201113235130-A
Country	US
Kind code	B2
Filing date	Sep 16, 2011
Priority date	Sep 17, 2010
Publication date	Feb 27, 2018
Grant date	Feb 27, 2018

Abstract

A system and method can perform subnet initialization in a middleware machine environment. In accordance with one embodiment, a plurality of management nodes in the middleware machine environment can be interconnected so that a plurality of subnet managers that are associated with the plurality of management nodes can communicate with each other using an in-band communication protocol. The plurality of subnet managers can negotiate and elect a master subnet manager, which operates to configure and manage the middleware machine environment. The master subnet manager can replicate one or more subnet initialization policies to other subnet managers in the middleware machine environment using the in-band communication protocol. The master subnet manager can then set up a plurality of client nodes in the middleware machine environment based on the one or more subnet initialization policies.

First claim

What is claimed is:

1. A system for performing subnet initialization in a middleware machine environment, the system comprising:
a plurality of network switches, wherein each network switch comprises a microprocessor executing a management node, wherein each management node comprises one policy daemon and one subnet manager; and
a plurality of client nodes interconnected by said plurality of network switches in a subnet,
wherein the plurality of subnet managers are in communication with each other using an in-band communication protocol in the subnet, wherein a first subnet manager of the plurality of subnet managers is an initial master subnet manager,
wherein the plurality of network switches have a first partition policy specifying a first partition membership of the client nodes in the middleware machine environment,
wherein the plurality of subnet managers negotiate with each other to elect an elected master subnet manager,
wherein a fail-over or hand-over of mastership occurs resulting in two or more master subnet manager instances, one master subnet manager instance of the two or more master subnet manager instances being the initial master subnet manager and another master subnet manager instance of the two or more master subnet manager instances being the elected master subnet manager,
wherein the elected master subnet manager uses the in-band communication protocol within a context of an initialization policy transaction to receive a new subnet initialization policy, wherein the new subnet initialization policy comprises a second partition policy different than the first partition policy, and wherein the second partition policy specifies a second partition membership of the client nodes in the middleware machine environment different than the first partition membership,
wherein the first partition membership of the client nodes in the middleware machine environment remains unchanged until the second partition policy is fully synchronized to the plurality of policy daemons,
wherein the elected master subnet manager replicates the new subnet initialization policy for the other subnet managers in the middleware machine environment as a replicated new subnet initialization policy,
wherein the elected master subnet manager, using the in-band communication protocol, distributes the replicated new subnet initialization policy to its respective management nodes for use by the subnet managers of each respective management node,
wherein the plurality of policy daemons obtain the replicated new subnet initialization policy from the respective subnet manager of its respective management node,
wherein the elected master subnet manager introduces one or more partition policy intermediate states to commit the change of the first partition membership of the client nodes in the middleware machine specified by the first partition policy to the second partition membership of the client nodes in the middleware machine specified by the second partition policy, wherein the one or more partition policy intermediate states are only either a subset of the first partition policy specifying the first partition membership of the client nodes in the middleware machine or a subset of the second partition policy specifying the second partition membership of the client nodes in the middleware machine,
wherein the elected master subnet manager commits the change from the first partition membership specified by the first partition policy to the second partition membership specified by the second partition policy in the two or more partial steps by reaching a number of the partition policy intermediate states where a quorum or majority of the plurality of subnet managers are updated with the second partition policy,
wherein the elected master subnet manager, after the second partition policy is fully synchronized to the plurality of policy daemons, updates the plurality of subnet managers to the second partition policy using the plurality of policy daemons and configures the plurality of client nodes in the middleware machine environment based on the second partition policy and the second partition membership.

2. The system according to claim 1 , wherein:
the plurality of client nodes are interconnected by the plurality of network switches in an Infiniband (IB) subnet.

3. The system according to claim 1 , further comprising:
a separate storage system that connects with the plurality of management nodes.

4. The system according to claim 1 , further comprising:
a plurality of host servers that connect with the plurality of management nodes.

5. The system according to claim 1 , further comprising:
one or more gateway instances that are selectively accessed by an associated guest.

6. The system according to claim 1 , wherein:
the plurality of policy daemons communicate using Internet Protocol (IP) over InfiniBand (IB); and
synchronization of the second partition policy is performed using an IB fabric.

7. The system according to claim 1 , further comprising:
a policy daemon collocated with the elected master subnet manager, wherein the policy daemon collocated with the elected master subnet manager configures the middleware machine environment using one or more policies.

8. The system according to claim 1 , wherein:
the middleware machine environment comprises an Ethernet network; and
the in-band communication provides communication redundancy in addition to Ethernet networking in the Ethernet network.

9. The system according to claim 1 , wherein:
the first and second partition policies define, respectively, first and second partition configurations in the middleware machine environment; and
the first and second partition policies are supplied to the middleware machine environment through an initialization policy transaction.

10. The system according to claim 1 , wherein:
the elected master subnet manager uses a default partitioning policy for initialization when the second partitioning policy is not specified by the new subnet initialization policy received by the elected master subnet manager within the context of the initialization policy transaction.

11. The system according to claim 10 , wherein:
the elected master subnet manager operates to configure one or more end nodes in the subnet as only limited members of a default partition based on the default partitioning policy.

12. The system according to claim 11 , wherein:
the elected master subnet manager operates to prevent illegal communication between said one or more end nodes which are only limited members of the default partition.

13. The system according to claim 1 , wherein:
the elected master subnet manager operates to synchronize fabric level configuration policy by starting up initially using a default partition policy, and wherein the elected master subnet manager proceeds to install a partition configuration in the middleware machine environment, when fabric level synchronization has been achieved.

14. The system according to claim 1 , wherein:
a replication and synchronization of the one or more partition policies is implemented in the two or more partial steps by the two or more instances of the elected master subnet manager as a result of the fail-over or of the hand-over of mastership.

15. The system according to claim 1 , further comprising:
a new master subnet manager different than the elected master subnet manager,
wherein the new master subnet manager dynamically discovers the second partition policy of the subnet and a current state of the subnet having the second partition policy;
wherein the elected master subnet manage

Assignees

• Johnsen Bjorn Dag
• Holen Line
• Arntsen Roy
• Oracle Int Corp

Inventors

• Johnsen Bjorn Dag
• Holen Line
• Arntsen Roy

Classifications

• H04L12/462

  LAN interconnection over a bridge based backbone · CPC title

• H04L69/40

  for recovering from a failure of a protocol instance or entity, e.g. service redundancy protocols, protocol state redundancy or protocol service redirection (management of faults, events, alarms or notifications in data switching networks H04L41/06) · CPC title

• H04L45/50

  using label swapping, e.g. multi-protocol label switch [MPLS] · CPC title

• H04L45/00 Primary

  Routing or path finding of packets in data switching networks (routing or path finding in wireless networks H04W40/00) · CPC title

• H04L41/0659

  by isolating or reconfiguring faulty entities · CPC title