Kefar | Uninet da la bienvenida a nuestro expositor de hoy |
Kefar | el Sr. Guido Fortunati, |
Kefar | residente en Irvine, California |
Kefar | se desempeña actualmente como administrador de red y |
sistema |
Kefar | La conferencia se titula: DoS Attacks |
Kefar | se solicitan traductores para esta conferencia, los voluntarios |
pueden intervenir |
Kefar | en el canal #redes |
Kefar | Adelante zuez |
Kefar | |
Kefar | |
zuez | hello everyone and welcome to the DoS tutorial |
zuez | this session has been created to make your internet life more |
enjoyable ;) |
zuez | topics to be covered today include: |
zuez | 1) modern denial of service |
zuez | 2) modern attacks |
zuez | 3) high rate attacks |
zuez | 4) attacking infrastructures :) |
zuez | 5) lan attacks |
zuez | 6) DDoS |
zuez | 8) How to filter with FreeBVSD |
zuez | FreeBSD even |
zuez | 10) protectiong router interfaces |
zuez | and last |
zuez | nuke information for all the windows population and tipos :) |
error: Connection reset by peer) |
zuez | my first topic will cover the ICMP type nuke.. |
zuez | lets say ssping or jolt |
zuez | old school programs.. |
zuez | these programs, send large and jumbled packets to windows |
boxes. |
zuez | any windows, wether you are running 9x, 98, Me, NT, whatever |
zuez | as a sidenote, jolt is not to be mistaken with the beverage :) |
#linux |
zuez | so anyways |
zuez | when windows attemps to re-asemble these packets into |
usable data,m it cant and your connection often slows down or just dies |
zuez | keep in mind, that while this is a pain, a simple reboot would |
fix it :) |
peter111[169.158.160.148]) |
zuez | i4d also would like to point out the main differente between |
ssping and jolt |
zuez | first up, jolt.. |
faiku[212.174.49.117]) |
zuez | jolt sends large and oversized packets top windoiws boxes |
zuez | sorry, windows boxes. |
zuez | ssping sends smaller packets, so that4s pretty much the main |
difference.. |
zuez | now |
zuez | first |
zuez | winnuke is a straight tcp pagket, sent to port 139 |
zuez | common netbios/samba port.. |
zuez | can you follow me, vixard? |
zuez | ok anyways, folks.. |
zuez | winnuke will send you OOB packets |
(Read error: Connection reset by peer) |
zuez | and no, OOB is NOT out of bound |
zuez | keep that in mind |
zuez | OOB stuff can also be send to other ports.. |
zuez | so to speak, OOB won4t break things. |
zuez | actually, that4s why most port blockers work for windows |
boxes. |
zuez | now, the other one i will explain before DoS stuff is ssping |
zuez | ssping works with the icmp protocol.. |
zuez | 1) its od, and incepted only for LANs. |
zuez | 2) it4s easy to spoof, and no, there4s no way to trace spoofs :) |
zuez | 3) its implementation doesn4t even seem to be the same for |
BSD/Linux, etc |
Connection reset by peer) |
zuez | now, windows, being a little bit silly as many of you know :) |
zuez | expects that when it gets an icmp packet its going to be atleast |
64bytes |
zuez | so, what ssping do? |
#linux |
zuez | sends 63 bytes :) |
#linux |
#linux |
zuez | so when windows tries to unravel and decipher the packet it |
comes up to one byte short, and chokes on a icmp furball |
zuez | this is as deep as i am going with this matter |
zuez | ask your questions in #qc :) |
zuez | ok, i am about to end with this matter.. |
zuez | to put it simply, two words, memori write :) |
off) |
zuez | when windows expects to write something of a certain size |
and it cant, it will overwrite memory, move memory around, and eventually |
choke |
zuez | #qc <MJesus> <viXard> que es "jolt" ??? |
zuez | let me explain what is jolt once again |
zuez | jolt sends large and oversized packets like i said before. |
zuez | just to confuse your windows box |
zuez | this is as deep as i am covering this guys, if someone has any |
kind of issue please ask. |
#linux |
zuez | ok |
zuez | sorry for the delay |
zuez | macs shouldnt be affected |
zuez | and no UNIX systems are affected by this. |
zuez | so, if you run UNIX/MacOS you don4t have to worry about this. |
zuez | of course there are many ways to bring a mac down his |
knees, but i won4t give out the codes :P |
zuez | ok |
zuez | i will be explaining now modern denial of service attacks.. |
zuez | oh |
zuez | sorry |
zuez | i skeeped something elite |
zuez | no, nevermind. |
zuez | ok |
zuez | first off, serious modern Denial of Service attack looks nothing |
like traditional attacks |
Pabli[trinidad.ssp.sld.cu]) |
zuez | #redes <viXard> espera a que el paquete icmp sea de 654 |
bytes |
zuez | the modern attacks which destroy services generally fall into |
one or more of the following categories: |
zuez | high rate floods |
zuez | ansy ffflood of packets which is not designed to waste |
bandwidth, but instead is designed to waste CPU and processing abilities, |
can be quite devastating |
zuez | any, sorry |
zuez | in this case, syn floods.. |
zuez | the evolution of the SYN flood has brought about the separate |
evolution of the high rate flood, which now has a life of its own |
zuez | now, let me explain real quick how to protect your BOX from syn |
attacks if you run FreeBSD |
zuez | you can simply add this line to your kernel, it should help |
zuez | options TCP_DROP_SYNFIN |
zuez | beyond that, its a matter of timme and CPU |
zuez | so, be aware. |
zuez | the second class is, infraestructure attack |
zuez | for well defended victims, it may be easier for the attacker to go |
after the network rather then sending packets directly to the true target |
zuez | now |
zuez | DDoS folks |
zuez | DDoS is much more than just multiplication of attack sources |
zuez | it brings about issues of path diversity, obscurity, invisibility, and |
demoralization of the victim. |
zuez | word |
zuez | ok |
zuez | any question? |
zuez | next topic will be high rate floods |
* zuez takes a 30s break :-) |
zuez | ok |
> thanks, the translator are too busy, zuez |
#linux Cannot send to channel |
Kefar | ok zuez |
zuez | #qc <Ston> es winxp vulnerable a algunos de los DoS |
mencionado anteriormente ? |
zuez | yes, certainly it is |
zuez | i will dcc you a reg patch after this session if thats ok with you. |
zuez | yo t enviare un parche luego de la conversacion te enviare un |
parche para el XP ston |
zuez | ok |
zuez | when i say high rate floods, i mean SYN floods |
zuez | believe me, SYN flood can be quite devastating.. |
zuez | i4ve had to drive all over SFO to SD to unplug networks just to |
stop SYN attacks, you cannot do too much about it.. |
zuez | the original goal of the SYN flood was to overwhelm a small |
queue of outstanding half-open connections with a very small amount of |
bandwidth |
zuez | so, in other words, this will prevent new connections and use |
your CPU |
zuez | specially if you are kinda poor and run a pentium 100MHz like i |
do, its painfull! |
zuez | besides, this attack is against the tcp implementation |
zuez | it can be used from a dialup connection to bring T1s/T3s and |
larger bandwidth to their knees |
zuez | #qc <sarnold> I'll explain why that opinion is *Very* stupid after |
zeuz is finished |
zuez | any issue peep? |
viXard | maybe hi´s talking about my paste |
zuez | sorry for the delay, i am waiting for sarnold |
zuez | alright |
zuez | ok |
zuez | one way to fix the SYN problem was to develop a better TCP |
stack |
zuez | just like fbsd did and linux.. |
zuez | (for all you linux lovers down here!) |
zuez | i think they used hash tables to do this, but maybe people like |
HoraPe, cron or vizard can help you with this better than me. |
zuez | so anyways |
zuez | the concept of the syn cookie was introduced essentially |
encoding all state information necessary for the connection to be opened in |
the return SYN|ACK |
zuez | so, no state needs to be mantained in the victim machine.. |
zuez | now |
zuez | as far as i know, some time around mid 1998 |
zuez | syn just came back to the itnernet life.. |
zuez | this time the goal wasn't simply to overflow a queue and prevent |
new connections, it was to generate packets so fast that the victim spent all |
their time processing them |
zuez | to put it simply, CPU killer! |
zuez | it dont matter how cool your box is, bloiieve me |
zuez | i have had this PIII 800MHz with 1Gb sitting behind cisco pix and |
checkpoints and someone killed our whole network when i ran an IRCtoo IRC |
server :) |
zuez | the high rate flood highlights the problems of existing tcp/I`p |
implementations |
zuez | tcp/ip sorry |
zuez | the amount of overhead which goes into handling each frame, |
inspecting each header, and processing each packet is large |
zuez | so, figure yourself :) |
zuez | when doing packet/sec calculations, remember that link layer |
overhead starts to play a major factor |
zuez | let me find something real quick for you folks |
zuez | ok here it is |
zuez | it works something like this |
zuez | Preamble and SFD (8 bytes) |
zuez | + Ethernet Header (14 bytes) |
zuez | + Payload (40 bytes in a SYN flood) |
zuez | + Frame Padding (6 bytes) |
zuez | + Frame Checksum (4 bytes) |
zuez | + Inter Frame Gap (12 bytes) |
zuez | #qc <sarnold> zeuz, was that a problem with the FreeBSD stack, |
with a very long execution path for any given packet? |
zuez | yes, it has been fixed with the 4.x/5.x branch, altho there are still |
many issues about this i will cover later on if you don4t mind, for all you |
FearBSD guys you can go to www.freebsd.org, however, i will explain how the |
FreeBSD stack works l |
zuez | so, your best way is, if you are running 3.x which is obsolete, |
cvsup to 4.x. |
zuez | so.. |
zuez | #qc <peter111> yo quiero pregutar algo sobre la conferencia... |
zuez | dime |
zuez | #qc <peter111> quien sabe donde conseguir un programa para |
Atackar por MS-DOS desde Win98 S.E |
zuez | you dont, or i am reporting you to the FBI autorities. |
zuez | so |
zuez | lets go on with this folks |
zuez | remember, attacking networks is bad. |
zuez | for example a 10Mbps ethernet pipe will max out at less then |
5Mbps of IP, when handling smallest-size packets. |
zuez | please, Horape or riel or anyone correct me if thats wrong |
please. |
zuez | now |
zuez | lets go on with secton 4 folks |
zuez | attacking the infraestructure of a network.. |
zuez | the design of most routers involves a central processor which |
handles routing protocols and administrative functions |
zuez | wether ytou have a cisco, juniper, whatever, it has a processor, |
ram, etc |
zuez | its pretty much a computer designed to do networking duties |
zuez | the traditional design of routers placed certain exceptional |
packets on the slow path, which requires attention from that processor |
zuez | so |
zuez | keep in mind, any time your router checks for a path or anything, |
it will eventually use the processor |
zuez | and yes, to avoid the questions, it can easily grash with a big |
attack |
zuez | however, i do believe cisco 12000 series, use distributed |
processors on each line card to handle the majority of routing without touching |
the main route processor used for routing protocols |
zuez | never used a cisco 12000 of course .) |
zuez | just general knowledge ;) |
zuez | when the processor which handles administrative and routing |
functions handles any packets at all, and particularly when it lacks good |
scheduling functions, it becomes vulnerable to denial of service |
zuez | i think you can get individual cards for cisco to continue |
forwarding packets |
zuez | i am not sure, i am sorry :) |
zuez | so, what happens if you are running an ISP and you have a |
strong attack folks? |
zuez | thanks horacio pena (horape) for the cisco information :) |
zuez | he just told me cisco can handle that nice feature, with few bugs |
tho.. |
zuez | so, lets go on with this folks.. |
zuez | so, if you happen to experience a big attack, you may see the |
router may not even be responsive at the local console, as the CPU spends its |
time processing interrupts and packets |
zuez | so, dont try to do anything with it, just unplug the cord and wait |
for a stop please. |
zuez | in the other side, attacks which overwhelm the route processors |
can be particularly bad when BGP is disrupted |
zuez | If a BGP speaking router is held down long enough for its peers |
to time out the keep alive and tear down the session, the routes get withdrawn.. |
zuez | and, plus.. |
zuez | If this removes the route used to carry the attack, the victim |
becomes unreachable and the attack is discarded further upstream |
zuez | that should bring your routerto l life too |
zuez | to life sorry :) |
zuez | but, as soon as the peer is established, the attack will begin again |
#linux |
zuez | (hope this explains a little bit why i dont want you to learn DoS |
skills, peter ;) |
zuez | ok |
zuez | lets talk about juniper routes now |
zuez | i am not that familiar with juniper, in fact, i know a little bit about |
them as i have had to deal with FreeBSD+juniper :) |
zuez | all i know is that juniper fare much better against this kind of |
attack because of their clean separation between packet processing and the |
routing engine |
zuez | thats whu i like junipers :) |
zuez | why. |
zuez | even exceptional packets which cannot be handled by an ASIC |
have a dedicated processor which limits the destructive potential of this kind of |
attack |
zuez | sorry for the delay, i am reading #qc as well :) |
zuez | #qc <velco> hmm, got a DDoS idea, instead DDoSing the |
original target, original target's defense measures, e.g. cut the |
zuez | routes, thus isolating innocent r00ted hosts which carry the |
attack. |
zuez | you dont |
zuez | velco, you dont want to kill your upstream and get them to |
discard your packets buddy. |
zuez | now.. |
zuez | the most common way to attack a router is to send packets |
destined to one of its local interfaces |
zuez | wether its a cisco, juniper, diet coke router, whatever, thats the |
easier way to attack a router.. |
zuez | some folks are msging me asking for smurf attacks, i will cover it |
later on too.. |
zuez | sorry if i skeiped smurf attacks |
zuez | so, lets keep talking about outers |
zuez | a cisco grp can be crippled by as few as 20,000 packets per |
second or so, i don4t know exactly (i am not a machine, again folks, horape |
riel or any folks probably know more about routers than i do..) |
timeout) |
zuez | the funny part is .. |
zuez | syn ports cannot only be sent to open ports on roputers like |
telnet.. |
zuez | but floods to random ports can be way too more devastating |
zuez | another way to generate exceptional packets is the use of IP |
options |
zuez | like hrm |
* zuez remmebers ;) |
zuez | router caches.. |
zuez | the most damaging thing of a SYN flood is the rst or ack replies |
generated in response |
zuez | dont worry tho, i will explain how to protect random routers |
against this problems :-) |
zuez | so now |
packets/sec can also be generated in a twist in the usual use of a smurf attack |
zuez | how does it work? |
zuez | using the network broadcast.. |
zuez | it will try to generate router harming effects instead of large |
packets designed to use large amount of bandwidth |
zuez | once, my friend ramiro (zero) tried to smurf his neighbor |
because he was using too much bandwidth (keep in mind they share a aDSL |
line) and he got smacked down with a 4pounds hammer |
zuez | j/k. |
zuez | anyways |
zuez | the attack of choice is still syn |
zuez | any qurestion? |
zuez | err, any question? |
zuez | ask your questions in #qc. |
zuez | ok |
zuez | now i am covering LAN attacks |
zuez | mainly because my friend asked |
#linux |
zuez | a variant of the smurf broadcast flood not commonly considered |
is the link layer broadcast flood |
zuez | in an atTack like smurff.. |
zuez | attack. |
zuez | packets are directed at an IP broadcast address, and the |
router/gateway will convert the packet into a layer link broadcast.. |
zuez | so |
zuez | one way to prevent yourself frm people like ramiro is to idisable |
directed broadcast request.s |
zuez | requests.. |
zuez | however, the attacker shall cause a LAN attack/smurf if he is in |
the same broadcast |
zuez | just like the example i gave you folks |
zuez | ramiro was using the same broadcast his neighbor does.. |
zuez | its also posible to generate a raw frame and forge the source |
macaddress to make the attack more dificult to trace.. |
zuez | anyways |
zuez | you can get good switches that can distinguish between |
broadcast and multicast traffic |
zuez | another potential area for LAN DoS is a spoofed ICMPs.. |
zuez | eek, ICMP |
zuez | err |
zuez | ICMP redirected to ARP :) |
zuez | thats it! |
zuez | which, will trick traffic into taking a detour ;) |
zuez | this not only create DoS, it can also be used to redirect trafic to |
another network for sniffing and such things |
zuez | any question? |
zuez | ok |
zuez | i already explained DDoS a little bit |
zuez | and i know many of you are a little bite tired/bored .. |
zuez | #qc <Ston> el ataque OOB fue resuelto en win98 es cierto que |
volvio a la luz en winxp ? |
Questions and Commentary at #qc' |
zuez | no. |
zuez | 2000/XP NO son vulnerables. |
zuez | next chapter |
zuez | How to filter? |
zuez | filtering smurf is quite easy, as you dont need ICMP echo replies.. |
zuez | let me cover how to filtetr w/cisco routers.. |
zuez | no service tcp-small-servers to prevent abuse of the small |
services for DOS or other attacks |
zuez | so thats tcp-small-servers |
zuez | you can asl |
zuez | err |
zuez | silly keyboard |
zuez | you can also add no service udp-small-servers |
zuez | ip route 0.0.0.0.0.0.0.0 null 0 253 todiscard packets with invalid |
destination addresses |
zuez | err |
zuez | error |
zuez | that should be ip route 0.0.0.0.0.0.0.0 null 0 255 folks |
peter111[169.158.160.151]: Connection reset by peer) |
zuez | thats as dee as i am going with cisco, if you need furter |
assistance feel free to msg me after this session or visit www.cisco.com |
zuez | ip route 0.0.0.0 0.0.0.0 null 0 255 |
zuez | thanks horape :) |
zuez | ok |
zuez | i think this session is over folks |
zuez | its taking way too much time |
viXard | :) |
zuez | does anyone have any question regarding anything? |
viXard | time to ask folks |
viXard | #qc for it |
zuez | well |
zuez | take what you have learned and use it to help yourself and |
others whenever possible |
viXard | nice explanation ;) |
> clap clap clap clap clap clap clap clap clap clap |
> clap clap clap clap clap clap clap clap clap clap |
> clap clap clap clap clap clap clap clap clap clap |
> clap clap clap clap clap clap clap clap clap clap |
> clap clap clap clap clap clap clap clap clap clap |
> clap clap clap clap clap clap clap clap clap clap |
Kefar | plas plas plas |
Kefar | plas plas plas |
Kefar | plas plas plas |
Neo | ;))) |
Kefar | plas plas plas |
zuez | :) |
> clap clap clap clap clap clap clap clap clap clap |
> clap clap clap clap clap clap clap clap clap clap |
> clap clap clap clap clap clap clap clap clap clap |
zuez | thank you folks. |
> clap clap clap clap clap clap clap clap clap clap |
Neo | zuez well done ;) |
> clap clap clap clap clap clap clap clap clap clap |
> clap clap clap clap clap clap clap clap clap clap |
MySQL | bravo!!!!!!!!!!!!!!!!!!!!! |
Zero | clap clap clap clap clap clap clap clap clap clap clap clap |
dani | plas plas plas plas plas plas plas plas <-- plas Flood DoS Atack |
dani | plas plas plas plas plas plas plas plas <-- plas Flood DoS Atack |
dani | plas plas plas plas plas plas plas plas <-- plas Flood DoS Atack |
dani | plas plas plas plas plas plas plas plas <-- plas Flood DoS Atack |
dani | plas plas plas plas plas plas plas plas <-- plas Flood DoS Atack |
peter111 | plas plas plas plas plas plas plas plas plas |
peter111 | plas plas plas plas plas plas plas plas plas |
peter111 | plas plas plas plas plas plas plas plas plas |
> clap clap clap clap clap clap clap clap clap clap |
> clap clap clap clap clap clap clap clap clap clap |
> clap clap clap clap clap clap clap clap clap clap |
zuez | i would like to thank horape, mjesus, kefar, riel, cron among |
others to the great support, and Zero for being a golden boy. |
elzo | plas plas plas plas |
elzo | plas plas plas plas |
> clap clap clap clap clap clap clap clap clap clap |
peter111 | Otro!!!! Otro!!!! Otro!!!! Otro!!! |
> clap clap clap clap clap clap clap clap clap clap |
MySQL | plas plas plas plas plas plas plas |
> clap clap clap clap clap clap clap clap clap clap |
MySQL | plas plas plas plas plas plas plas |
MySQL | plas plas plas plas plas plas plas |
> clap clap clap clap clap clap clap clap clap clap |
MySQL | plas plas plas plas plas plas plas |
Zero | lol |
> clap clap clap clap clap clap clap clap clap clap |
> clap clap clap clap clap clap clap clap clap clap |
* cron felicita a zuez por el trabajo |
MySQL | bravo!! |
peter111 | clap clap... clap |
peter111 | clap clap... clap |
MySQL | bravo!! |
peter111 | clap clap... clap |
MySQL | torero! |
MySQL | torero! |
peter111 | plas plas plas |
peter111 | plas plas plas |
peter111 | plas plas plas |
zuez | muchas gracias :) |
viXard | <Ston> es posible detectar un DoS que se produsca dentro de |
una red... si es posible como se arrria? |
* peter111 le gusto mucho la conf. |
viXard | <Ston> haria |
Ston | s/arrria/haria |
zuez | ok |
clap clap clap clap clap clap clap clap clap clap clap clap clap ' |
peter111 | Felicidades zuez |
> clap clap clap clap clap clap clap clap clap clap |
> clap clap clap clap clap clap clap clap clap clap |
zuez | ston, what OS? |
zuez | anyways, if you happen to run snmp you should see a big traffic |
going through your interfaces :) |
Ston | Linux To Win* |
zuez | ston: tcpdump :) |
JALH | clap clap clap clap!! |
peter111 | MJesus: cuando es la proxima conferencia? |
JALH | :) |
> :)) |
Ston | ahhhh the admin use winnt and i use linux... |
zuez | Ston: run any application that lets you debug your itnernal traffic. |
dardhal | nice lecture, very good |
zuez | i dont tend to run NT, i dont know :( |
Ston | ok |
zuez | thank you. |
viXard | peter111: el lunes |
> Dec, 9. 22:00 |
> Horst von Brandt. (Chile) |
> title pending ! |
sarnold | zuez, thank you :) I'm sorry I phrased my message to viXard |
so poorly .. |
viXard | http://grc.com/dos/ |
Ston | es posible para totalmente un DoS ? |
HoraPe | viXard, ugh! |
zuez | sarnold, no problem man |
zuez | Ston: no, de hecho no hay mucho que puedas hacer :) |
cron | cosas como "echo 2 > /proc/sys/net/ipv4/tcp_keepalive_probes" |
y "echo 30 > /proc/sys/net/ipv4/tcp_keepalive_time" pueden ser util. Ref: |
http://www.linuxdoc.org/LDP/solrhe/Securing-Optimizing-Linux-RH-Edition-v1.3 |
/chap6sec75.html |
Exiting) |
viXard | HoraPE: huh ? |
HoraPe | que esa página no vale nada... |
zuez | me voy llendo |
zuez | que tengan un muy buen fin de semana. |
viXard | no ? |
botijo | no es grave, Jorge...lo grave es el complejo de logo |
HoraPe | el que la escribe es un idiota |
dani | zuez: gracias por la charla ;) |
viXard | bueh |
zuez | gracias :) |
> graciassssssssssssssssssss |
* zuez se fue. |
viXard | corre que te corre XD |
Ston | ping -f -s -l 65510 zuez |
Ston | ;) |
Ston | yo me voy para mi casa, dejo de llover y la oficina da asco |
Ston | adios |
here somewhere...) |
sarnold | vixard; grc.com's frequent ranting about WindowsXP is |
completely unfounded. |
sarnold | vixard: as i understand grc.com's complains against WinXP, it |
is primarily the 'raw sockets' that he doesn't like |
viXard | tell me more |
sarnold | vixard: however, it completely escapes him that 4.2BSD |
introduced sockets back in 1983... |
viXard | jej |
sarnold | vixard: Completely free *BSD systems have had this capability |
since 1994 at the absolute latest |
viXard | well, i just read the story about their DoS attack |
sarnold | vixard: and, I'm reasonably certain Linux had the same |
capabiltiies around 1993, maybe earlier, again -- completely free |
End of #linux buffer Fri Dec 07 23:23:38 2001 |