Home NetBrain Product One-Skill-at-a-Time Search Mark Forums Read
Go BackThe Net-BrainerNetBrain Product How does NetBrain discover network?

Thread Tools Search this Thread Display Modes
Old 11-20-2009, 03:38 PM
read2success read2success is offline
Join Date: Nov 2009
Location: Boston
Posts: 2
Default How does NetBrain discover network?

Our network has thoussands of routers, switches, and firewalls from Cisco, Juniper, Checkpoint, F5... The discovery tools we tried failed badly.

What's different with NetBrain?
Reply With Quote
Old 11-20-2009, 04:39 PM
lgao lgao is offline
Join Date: May 2009
Posts: 34
Default Re: How does NetBrain discover network?

The common challenges to the discovery tools are
* Accuracy
Unless network is designed perfectly like in the lab, a round of discovery could find un-predictable percentage of devices

* Speed
Most tools that rely on network-scanning to discover will choke when network devices are in thousands, and end points are in tens of thousands. It could take a week to finish a round of discovery, if it finishes.

* Output
Early generation tools use SNMP/Ping scan to discover L3/L2 devices and the connectivity. The output would be a landry list of devices as long as hundreds of pages.

-----> The answer from NetBrain

A few years ago when a dedicated team from NetBrain starts to tackle the discovery for large enterprises, we set the following goals: 1) Bring accuracy to around 99% 2) full discovery needs to be done in a day, even for world's largest network (30k devices and millions of end points) 3)output needs to be map format, not CSV file.

So the difference of NetBrain discovery is that we are very close to all the above goals. Recently one of Fortune 50 companies used NetBrain for discovery and documentation: 20,000 routers, switches, firewalls and 0.5 million servers were discovered in 10 hours. Later on they verified the accuracy with manually-maintained record and claim it reached 98.7% accuracy. Most importantly to them, over 1000 map (l2 and L3 Visio format) were generated automatcally for the 650 data centers and sites.

---> The secret of NetBrain discovery

NetBrain used a set of compound alogrithm similarly to how human expert would do to discover an unknown network - called neighbor-walking mechanism. It dynamically analyze configure file, routing table, routing peer, CAM table, ARP table etc on top of traditional discovery mechanism to "figure out" what's in the network. It never relies on brute-force scanning. While discovering the devices, it then applies a decoding engine to dissect the design of network, and crank out annotated network diagrams as output.

The solutuion is packaged in two ways: for network smaller than 200 network devices, use the enterprise edition workstation; for larger network with tens of thousands of network devices, use discovery appliance.

Happy NetBraining...

NetBrain Team
Reply With Quote
Old 11-21-2009, 05:42 PM
lgao lgao is offline
Join Date: May 2009
Posts: 34
Default Re: How does NetBrain discover network?

* Why Neighbor Walking Can Discover Network Faster? *

Network, by definition, is connected somewhere somewhat. But unless the network is real simple, it is very hard to find "One Spot" that can reach everwhere - this is why brute force scanning not only take time and your network resource, it can not yield high accuracy.

Neighbor-walking is to ask friends about friends and friends' friends... until evey friend is counted. - It is much more difficult to write such program (an order of magnitude), but here is the benefit - it will be fast and accurate. Why?

Consider the following puzzle: if I give you a large piece of paper, and I ask you to fold it once, and twice, and third ... up to 50 times.
How tall do you think the final stack is going to be?

[Answer] the distance from Earth to Sun

--> The theory behind neighbor walking is the same.
Reply With Quote


Thread Tools Search this Thread
Search this Thread:

Advanced Search
Display Modes

Forum Jump

All times are GMT -4. The time now is 09:38 PM.
Powered by vBulletin
Copyright 2000-2010 Jelsoft Enterprises Limited.
Copyright 2009 NetBrain, Inc. All rights reserved.