Proactive steps to prevent DDOS?

Fine, does this work better for you?

Help me, what proactive steps can I take to protect my network from a DDOS?

There isn't a lot that can be done, but there are a few steps you can take to "get ready" for a DDOS attack.

   --Make sure you have monitoring of your routers or firewalls in place
     so you'll get an early alert of a possible DOS attack. This will at
     least allow you to start working on the problem (and drafting
     press releases :-).
    --Talk to all of your up stream providers so you know how to contact and
     work with them if they are a source of a DOS attack against you. If your
     up stream provider isn't willing to work with you on this, start the
     process of getting a new up stream provider.

   --Look into the systems that are being developed and starting to become
     available that help automate the work to diagnose DDOS attacks.
     Encourage your up streams to do the same.

   --Make sure you have in place the filtering on your own networks that you
     wish everyone else had in place on their networks. This won't protect
     you from being attacked, but it will prevent you and your users from
     attacking others (or at least using spoofed IP addresses to do so), and
     that in turn may prevent you from being the target of a retaliatory DOS
     attack. It can also prevent or limit the spread of a DOS attack that
     originates within your network or from someone down stream. From your
     customer's point of view there may not be much difference between
     you being the source of or the target of a DOS attack--either way
     performance is likely to be poor and customers are likely to be unhappy.

   -Jeff Ogden
    Merit

We've got this document on the company's public-facing web site:
http://www.genuity.com/securitymatters/rundown004.htm

and I think we've also got some more documents that our Customer Service Center distributes.
Have you tried calling Genuity and asking? I would expect that you would get a productive response.

cheers
Betsy
(not in that area of things myself)

Help me, what proactive steps can I take to protect my network from a DDOS?

There isn't a lot that can be done, but there are a few steps you can
take to "get ready" for a DDOS attack.

   --Make sure you have monitoring of your routers or firewalls in place
     so you'll get an early alert of a possible DOS attack. This will at
     least allow you to start working on the problem (and drafting
     press releases :-).

I would add careful use of some rate-limiting functionality,
(already mentioned in Richard Steenbergen's http://www.e-gerbil.net/ras/dos.txt)
so you can rate-limit things like icmp and acks numbered 0 and anything
else that show themselves to be obvious candidates over time.

As well, for us, part of "being ready" is knowing how to:
- deploy rate-limiters (so you're not trying to learn it during an attack)
- having ways to identify the target(s) of the attack -- predefined
   ACLs that are not applied but can be, to help determine the target(s),
   rmon probes, working knowledge of tcpdump or network sniffers, etc.

i.e. deploy ACLs on egress routers that can be enabled when an attack begins.
A set of carefully defined filters can help determine information about the
attack -- perhaps the src or dst or ports, or the fact that it's all random.
Then ACLs to block or rate-limit can be built and applied, and/or if you're
asking your upstream to take actions to help you, you can help them help you
if you have information about the nature of the attack.

Granted it's a blunt instrument, but it has worked for me in the past
when an attack has rendered better tools useless.

Fine, does this work better for you?

Help me, what proactive steps can I take to protect my network from a DDOS?

There isn't a lot that can be done, but there are a few steps you can take to "get ready" for a DDOS attack.

  --Make sure you have monitoring of your routers or firewalls in place
    so you'll get an early alert of a possible DOS attack. This will at
    least allow you to start working on the problem (and drafting
    press releases :-).
  --Talk to all of your up stream providers so you know how to contact and
    work with them if they are a source of a DOS attack against you. If your
    up stream provider isn't willing to work with you on this, start the
    process of getting a new up stream provider.

  --Look into the systems that are being developed and starting to become
    available that help automate the work to diagnose DDOS attacks.
    Encourage your up streams to do the same.

I know of just Asta Networks:
Asta Networks claims cure for denial-of-service attacks, Jan 17, 2001
http://www.nwfusion.com/news/2001/0117ddos.html
Firm eyes DOS attacks, Jan 22, 2001
http://www.nwfusion.com/archive/2001/115979_01-22-2001.html

Can you elaborate on others you may know?

-Hank

[snip]

Does anyone have any nifty ideas on how we can *squash* this?
The script kiddies are only going to get more clever (or their
utilities are), and it seems that the majority of the ideas
people have for DoS/DDoS seem to be limited to trying to
treat the symptoms.

(.. which, even demonstrated yesterday by someone hosting an efnet
server, requires extremely good coordination and contacts with
the "large players" for it to be effective.)

2c,

Adrian

Yes, Asta is one.

There is a DARPA funded research project called Lighthouse at the University of Michigan that is working in this area. Merit has been involved mostly by giving them access to traffic on a real operational network. See:

      http://www.darpa.mil/leaving.asp?url=http://www.eecs.umich.edu/lighthouse

I understand that there are other DARPA funded efforts working on different aspects of the DOS problem (automatic detection, trace back, counter measures).

Take a look at "Networking & Distributed Systems" under

      http://www.darpa.mil/ito/ResearchAreas.html

In particular see:

      http://www.darpa.mil/ito/psum2000/J032-0.html
      http://www.darpa.mil/ito/psum2000/J910-0.html
      http://www.darpa.mil/ito/psum2000/J028-0.html

DANTE has also developped a tool made of in-house scripts, a database and based on netflow exports, that detects more DoS attacks than manpower is available to treat.

Still, it enables us to log, and treat, the major (long lasting, repeting, extremely distributed, powerful, you name it) ones.

However, we have discovered the following interesting paradox:
- the most transit traffic a network carries, the most likely it will also carry DoS attacks, the most DoS attacks will be noticed and the higher the costs associated to DDoS will be
- once an attack is detected on a transit network, getting the correct administration of the end sites to actually do something about it, is the real problem, especially if those end sites are not direct peers (which, for some major transit networks, is always the case).

As usual, it is enough one administration in the chain has not enough manpower/does not understand the problem or ways to fix it/thinks the problem is not worth fixing/has different priorities for DDoS compromised hosts to remain compromised for months.

Its good to see the awareness is being raised recently, though.

DH.

Any chance folks on this list would be willing to share the source (or
failing that, _detailed_ tech specs) to their DoS detection tools? Or
is the general consensus still that this is a big no-no, not even up
for discussion, as such tools are highly proprietary, and the public
scrutiny resulting from such a release would be a Bad Thing(TM)?

I'm sure I'm not alone in saying that lots of smaller folks wouldn't
mind deploying these, but lack the in-house resources needed to
develop them from the ground up.

-adam