Browse Prior Art Database

Method To Reduce Processing Of Redundant Events In Environments With Multiple Agents Reporting On A Fabric Disclosure Number: IPCOM000201833D
Publication Date: 2010-Nov-25
Document File: 6 page(s) / 36K

Publishing Venue

The Prior Art Database


Disclosed is a method to reduce sets of agents to monitor fabric events while simultaneously ensuring that the fabric events are monitored on a continuous basis. The approach expands upon existing ideas to determine the agent to be selected as delegate at a certain point in time and solves the problem of continuous monitoring of fabric events, even when using a reduced set of agents to monitor the events.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 22% of the total text.

Page 01 of 6

Method To Reduce Processing Of Redundant Events In Environments With Multiple Agents Reporting On A Fabric

Management applications that monitor storage area network (SAN) fabrics need to be able to monitor for fabric events that indicate some changes in the underlying fabric environment. Such events need to be reported and possibly require data collection from the fabrics to refresh the fabric information of which the applications are aware.

There may be multiple agents that can report on fabric events for a fabric. Hence, a single event can cause multiple agents to report the same event to the server. The server may not be able to recognize the duplicate events due to lack of sufficient information in the event. In the specific case of fabric events received there is no unique identifier for the event; therefore, one cannot determine for sure whether multiple event notifications really correspond to the same underlying event. In addition, notification to the server from agents for the same event can be staggered over time making it not possible for the server to determine that the notifications are for the exact same event.

Redundant notifications for the same event by multiple agents results in redundant processing by the server. Some system fabric discoveries and probes performed may lead to unnecessary load on the fabrics. This issue can lead to scalability problems in very large environments wherein thousands of agents may be present. In addition, since every agent in the environment is monitoring fabric for events, there is a processing overhead on the machines on which the agent is running.

Mechanisms such as event quiescing can be used to reduce processing in response to redundant events. However, there is still overhead on both server and agent if all agents report on fabric events for a fabric.

Since there are multiple agents capable of reporting on a fabric, it is sufficient to have one agent to monitor fabric events for a fabric at a point of time and have that agent send the event notifications to the server. One method is to determine equivalent subsets of agents and make assignments to minimize the number of agents. (1) In the context of that method, the ability to send notifications for fabric events can be considered a capability.

However, in the case of event notifications the server needs to attempt to have an agent delegated to notify events on a continuous basis so that the server reacts to fabric events that may occur at any point of time or, at a minimum, realize time periods when event notification does not exist and perform a fabric discovery/probe to refresh the information (due to lack of event notification ability during that period). In this context "delegated" refers to an agent that has been selected for event notifications for a fabric.


Page 02 of 6

This requires that the server:
• Be aware that the delegated agent is still operational and actively monitoring fabric events

• Recognize when ano...