AutoNOC 2.5 User Guide
Preface
Acknowledgements
System Requirements
Legal

Part 1 - Introduction
1.1 The Ideal Difference
1.2 Automated Operations
1.3 Services & Scaler
1.4 Acquisition Stacks
1.5 Portal Deployment
1.6 Discovery and Crawler
1.7 Monitoring Agents
1.8 Recoiling Database
1.9 Multiple Languages
1.10 Security

Part 2 - NOC Views
2.1 Investigate
2.2 Observe
2.3 Visualize
2.4 Alarms
2.5 Analyze
2.6 Design
2.7 Configure

Part 3 - Model Design
3.1 Object Model
3.2 Devices
3.3 Sets
3.4 Set Criteria
3.5 Probes
3.6 Logs & Events
3.7 Alarms
3.8 Actions
3.9 Reports
3.10 Users
3.11 Polling
3.12 Service Levels
3.13 Dependencies
3.14 Performance

Part 4 - Developer Features
4.1 Adding SNMP MIBs
4.2 Variables
4.3 OSP API
4.4 Probe Template
4.5 Log Template
4.6 Device Template
4.7 Interface Template
4.8 Rebranding

Part 5 - Troubleshooting
5.1 General Issues
5.2 Linux
5.3 Windows

Appendix
A.1 OSP API Functions
A.2 Variables
A.3 Object Reference

1.6 The Holy Grail of Discovery?
Network discovery is the proverbial "Holy Grail" of network operations. Everyone dreams of perfection in network discovery, but few have found it, and a perfect automated discovery solution for one person is not often the best for another, who only wants certain conceptually selected devices. Many people observe that all the time it takes to tune things to make the "perfect auto-discovery" it would be just as fast to enter the IP addresses by hand for exactly the devices needed to be monitored.

Indeed, in our experience, a simple user-defined, manually entered operations model has often proven to be the shortest and fastest path to model exactly what the user needs/wants to model. This problem of seeking the Holy Grail of Discovery, combined with pragmatic limitations of perfection in discovery has led us to take a more hybrid approach.

We wanted to give the user the ability to do network scans and add devices automatically, but had to deliver this in a very targeted manner so that the software only searches designated discovery zones of the network. Additionally, the software had to support the people who prefer to build their own models explicitly through AutoNOC's interpreter or via the web GUI as there are quite a few network managers who simply do not allow network discovery on their networks.

The result of these experiences is the current implementation of discovery in AutoNOC. There are two components of it. The first component is the Network object. This is the user designed object that specifies targeted ranges in the network that AutoNOC should search within, including which tickets should be used, etc. The second component is the background Network Crawler. The network crawler scurries around in the operations model comparing the model to the real network. If it detects any changes between the model and the network it will make them.

1.6.1 Targeted Discovery with the Network Object
When a new user starts the software up, they will be prompted with a dialog box that simplifies a default network object down to seed IP addresses and the related community and agent names. This dialog pre-configures a default starting Network Object.

Each Network Object has separate targeted network zones, methods, rules, community names, etc. Everything needed to discover a certain part of a network. The network objects periodically execute and scan their designated target zones. They can add devices and also remove devices that are no longer present.

The screenshot above shows the methods tab of the network object.

1.6.2 Network Crawler
The Network Crawler is disabled by default. Once enabled (it is enabled as a property of the Discovery service) the software will periodically patrol the network model and search for differences between the model and the actual network objects. If it finds differences between the real object and the model it will modify the model to update it with the new changes.

Note that the software also makes use of a Discovery Grace Period. The grace period determines how long a given difference must exist prior to making the change to the model. This allows the software a window to detect, for instance, down interfaces, prior to actually removing them from the model. The network crawler runs as a background low-impact process. Devices that are currently being controlled are identified by a star in the Design tab of the GUI. The software will crawl one device at a time.

(C) 2007 - All Rights Reserved - AutoNOC LLC
      Call me!