Skip to contentSkip to conversation

DDoS mitigation

LARTC devik 2 participants 5 messages Aug 25, 2003 – Aug 26, 2003

lore
Share & tools
lore

devik

author
#0 lore raw

Hi,

I got idea how to create anti-DDoS framework. I depicted it here: http://luxik.cdi.cz/~devik/qos/ddos-blackhole.htm
I'd appreciate opinions whether it could work. Please Cc me in replies.

Thanks,
-------------------------------
Martin Devera aka devik
Linux kernel QoS/HTB maintainer
http://luxik.cdi.cz/~devik/

_______________________________________________
LARTC mailing list / [email protected]
http://mailman.ds9a.nl/mailman/listinfo/lartc HOWTO: http://lartc.org/

#1 lore raw

For TCP that works. There are, however, UDP applications that are one-way (e.g. streaming video/audio). Many multicast applications are one-way.

Quoted 16 lines · devik

Hi,

I got idea how to create anti-DDoS framework. I depicted it here: http://luxik.cdi.cz/~devik/qos/ddos-blackhole.htm
I'd appreciate opinions whether it could work. Please Cc me in replies.

Thanks,
-------------------------------
Martin Devera aka devik
Linux kernel QoS/HTB maintainer
http://luxik.cdi.cz/~devik/

_______________________________________________
LARTC mailing list / [email protected]
http://mailman.ds9a.nl/mailman/listinfo/lartc HOWTO: http://lartc.org/

Signature · 3 lines
    Lawrence MacIntyre     865.574.8696     [email protected]
               Oak Ridge National Laboratory
High Performance Information Infrastructure Technology Group
author
#2 lore raw
In reply to devik

I would ignore multicast ane let it go thru as aparently regular dialup and ADSL users have no access to it. Thus I consider it to be more secured by ISPs. Streaming audio/video, is not there some "feedback" channel so that server knows when client is dead ? There should be something like it IMHO. Note that I'd could every packet going to host (ignoring tcp/udp and/or port difference).

Also thanks to Gerry to take it so seriously. I'm interested in result - especialy because I got the idea in night while being tortured by gnats ;-)

devik

Quoted 18 lines · Lawrence MacIntyre

For TCP that works. There are, however, UDP applications that are one-way (e.g. streaming video/audio). Many multicast applications are one-way.

Quoted 13 lines · devik

Hi,

I got idea how to create anti-DDoS framework. I depicted it here: http://luxik.cdi.cz/~devik/qos/ddos-blackhole.htm
I'd appreciate opinions whether it could work. Please Cc me in replies.

Thanks,
-------------------------------
Martin Devera aka devik
Linux kernel QoS/HTB maintainer
http://luxik.cdi.cz/~devik/

Quoted 8 lines
Quoted 3 lines

_______________________________________________
LARTC mailing list / [email protected]
http://mailman.ds9a.nl/mailman/listinfo/lartc HOWTO: http://lartc.org/

--
Lawrence MacIntyre 865.574.8696 [email protected]
Oak Ridge National Laboratory
High Performance Information Infrastructure Technology Group

_______________________________________________
LARTC mailing list / [email protected]
http://mailman.ds9a.nl/mailman/listinfo/lartc HOWTO: http://lartc.org/

#3 lore raw
In reply to devik

If you are using RTP for the session, you may also use RTCP as a back-channel for reception reports. However, this is not required.

Quoted 42 lines · devik

I would ignore multicast ane let it go thru as aparently regular dialup and ADSL users have no access to it. Thus I consider it to be more secured by ISPs. Streaming audio/video, is not there some "feedback" channel so that server knows when client is dead ? There should be something like it IMHO. Note that I'd could every packet going to host (ignoring tcp/udp and/or port difference).

Also thanks to Gerry to take it so seriously. I'm interested in result - especialy because I got the idea in night while being tortured by gnats ;-)

devik

Quoted 26 lines · Lawrence MacIntyre

For TCP that works. There are, however, UDP applications that are one-way (e.g. streaming video/audio). Many multicast applications are one-way.

Quoted 13 lines · devik

Hi,

I got idea how to create anti-DDoS framework. I depicted it here: http://luxik.cdi.cz/~devik/qos/ddos-blackhole.htm
I'd appreciate opinions whether it could work. Please Cc me in replies.

Thanks,
-------------------------------
Martin Devera aka devik
Linux kernel QoS/HTB maintainer
http://luxik.cdi.cz/~devik/

Quoted 3 lines

_______________________________________________
LARTC mailing list / [email protected]
http://mailman.ds9a.nl/mailman/listinfo/lartc HOWTO: http://lartc.org/

--
Lawrence MacIntyre 865.574.8696 [email protected]
Oak Ridge National Laboratory
High Performance Information Infrastructure Technology Group

Signature · 3 lines
    Lawrence MacIntyre     865.574.8696     [email protected]
               Oak Ridge National Laboratory
High Performance Information Infrastructure Technology Group
author
#4 lore raw
In reply to devik

Aha. Ok. I just read about RTP and they argue that either RTPC or other session management protocol should be used.
I'd guess that any sane protocol would use some kind of liveness reporting from users. If recieving app crashes then the server should recognize it in some reasonable time and stop sending RTP packets, am I right or do I miss something ?

thanks, devik

Quoted 11 lines · Lawrence MacIntyre

If you are using RTP for the session, you may also use RTCP as a back-channel for reception reports. However, this is not required.

Quoted 7 lines · devik

I would ignore multicast ane let it go thru as aparently regular dialup and ADSL users have no access to it. Thus I consider it to be more secured by ISPs. Streaming audio/video, is not there some "feedback" channel so that server knows when client is dead ? There should be something like it IMHO. Note that I'd could every packet going to host (ignoring tcp/udp and/or port difference).

_______________________________________________
LARTC mailing list / [email protected]
http://mailman.ds9a.nl/mailman/listinfo/lartc HOWTO: http://lartc.org/