One thing which is not so obvious is to reduce false positive.
This is hard when you have a mix of traffic profiles/patterns within
your network, with customers in differents domains (scientists,
financials, video addicted, torrent addicted, etc...) with different
bandwidth.
a)
Does anybody tried to separate ip range by traffic profile to apply
specific rule/profile per ip allocation?
puts all financials clients into range X/X and define rule Z
puts all scientists clients into range Y/Y and apply rule Q
etc....
Does this help ?
b)
One other method could be to classify customers by their bandwidth.
profile 1. from 10-100M
profile 2. 100-500M
profile 3. 500M-1000M
profile 4. >1000M
Like this you do not mix big BW with small BW customer, and do not get
alerted when client from profile 4 start to download at 1G.
Any experience ?
My guess is that solution b is better than a. Not so easy to classify
traffic pattern per group of client.
Thank, best regards.
- Marcel
On 13.08.2015 06:42, Ramy Hashish wrote:
Hello Fabien,
And why don't you use A10 for both detection and mitigation?
Thanks,
Ramy
On Wed, Aug 12, 2015 at 6:42 PM, Fabien Delmotte <fdelmot...@mac.com> wrote:
Hello
My 2 cents
You can use Wanguard for the detection and A10 for the mitigation, you
have just to play with the API.
Regards
Fabien
Le 12 août 2015 à 16:28, Ramy Hashish <ramy.ihash...@gmail.com> a écrit
:
Date: Tue, 11 Aug 2015 08:14:54 +0200
From: "marcel.durega...@yahoo.fr" <marcel.durega...@yahoo.fr>
To: nanog@nanog.org
Subject: Re: Experience on Wanguard for 'anti' DDOS solutions
Message-ID: <55c992de.3020...@yahoo.fr>
Content-Type: text/plain; charset=windows-1252; format=flowed
anybody from this impressive list ?:
https://www.andrisoft.com/company/customers
-- Marcel
Anybody here compared Wanguard's performance with the DDoS vendors in the
market (Arbor, Radware, NSFocus, A10, RioRey, Staminus, F5 ......)?
Another question, have anybody from the reviewers tested the false
positives of the box, or experienced any false positive incidents?
Thanks,
Ramy
