Root-cause analysis of network performance problems

• US 7,489,639 B2
• Filed: 03/23/2005
• Issued: 02/10/2009
• Est. Priority Date: 03/23/2005
• Status: Expired due to Fees

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Breitgand, David, Shehory, Onn, Yerushalmi, Elena, Henis, Ealan, Ladan-Mozes, Edya
• Primary Examiner(s): Harper; Kevin C.
• Assistant Examiner(s): WONG, XAVIER S
• Application Number:
  US11/088,054
• Publication Number:
  US 20060215564A1
• Time in Patent Office: 1,420 Days
• Field of Search: 370/241, 370/242, 709223-224, 714/1, 714/25, 714/48, 714/799
• US Class Current: 370/241
• CPC Class Codes: H04L 41/0631 using root cause analysis; using analysis of correlation between notifications, alarms or events based on decision criteria, e.g. hierarchy, tree or time analysis
• CPC Class Codes: H04L 41/12 Discovery or management of network topologies
• CPC Class Codes: H04L 41/18 Delegation of network management function, e.g. customer network management [CNM]
• CPC Class Codes: H04L 43/0811 by checking connectivity

Abstract

A method for diagnosing network performance problems includes providing a graph representing a topology of a network including first nodes that are classified as providers of a resource, second nodes that are classified as consumers of the resource, and third nodes that are classified as mediators intervening between the providers and the consumers. Responsively to a report of a performance problem encountered by a given consumer in the network, at least one input/output (I/O) path is traced through the graph connecting the given consumer to one or more of the providers. One or more problematic nodes are identified along the at least one I/O path as possible causes of the performance problem.

Claims

1. A method for diagnosing network performance problems, the method comprising:

providing a graph representing a topology of a network comprising first nodes that are classified as providers of a resource, second nodes that are classified as consumers of the resource, and third nodes that are classified as mediators intervening between the providers and the consumers;

receiving a report of a performance problem encountered by a given consumer in the network;

responsively to the report, tracing at least one input/output (I/O) path through the graph connecting the given consumer to one or more of the providers;

identifying one or more problematic nodes along the at least one I/O path as possible causes of the performance problem;

tracing a further I/O path through the graph connecting another consumer to the providers and crossing the at least one I/O path at one of the problematic nodes after identifying one or more problematic nodes along the at least one I/O path; and

identifying one or more further problematic nodes along the further I/O path as a possible cause of the performance problem,

wherein providing the graph comprises identifying physical and logical links between the nodes, and

wherein tracing the at least one I/O path comprises building the at least one I/O path in accordance with the logical links.

2. The method according to claim 1, wherein identifying the logical links comprises providing a logical mapping between the nodes, and wherein building the at least one I/O path comprises eliminating from the path the physical links that do not correspond to the logical mapping.

3. The method according to claim 1, wherein providing the graph comprises grouping a plurality of the nodes together into a super-node having an internal node structure, and wherein tracing the at least one I/O path comprises tracing the at least one I/O path through the super-node without reference to the internal node structure.

4. The method according to claim 1, wherein the first nodes comprises a storage area network (SAN), and wherein the resource comprises at least one of a storage capacity and a bandwidth.

5. The method according to claim 4, wherein the first nodes comprise storage servers having logical volumes associated therewith, and wherein receiving the report comprises receiving an indication of at least one of a capacity problem and a throughput problem in one of the logical volumes.

6. The method according to claim 4, wherein the first nodes comprise storage servers having logical volumes associated therewith, and wherein the second nodes comprise storage devices having logical units associated therewith, and wherein tracing the at least one I/O path comprises extending the at least one I/O path from one of the logical volumes to one or more of the logical units.

7. The method according to claim 1, wherein tracing the at least one I/O path comprises building a directed acyclic graph (DAG) within the topology of the network from the given consumer to the providers that are coupled by the topology to provide the resource to the given consumer.

8. The method according to claim 1, wherein identifying the one or more problematic nodes comprises monitoring performance metrics of the nodes in the network, and identifying a given node as problematic if one or more of the performance metrics of the given node are outside a predefined bound.

9. The method according to claim 1, wherein tracing the further I/O path comprises initiating the tracing of the further I/O path upon determining that the one or more problematic nodes identified along the at least one I/O path are not the causes of the performance problem.