[members-discuss] Technical solution to resolve Spoofed IP traffic, Spoofed amplification DDoS attacks, BGP&RIR hijacking, IoT botnet infections and Botnet C&Cs

Cynthia,

This is a presentation that I will show in the near General Meeting.

Respectfully,
Elad
________________________________
From: Cynthia Revström <me at cynthia.re>
Sent: Thursday, April 30, 2020 11:57 PM
To: Elad Cohen <elad at netstyle.io>
Cc: members-discuss at ripe.net <members-discuss at ripe.net>
Subject: Re: [members-discuss] Technical solution to resolve Spoofed IP traffic, Spoofed amplification DDoS attacks, BGP&RIR hijacking, IoT botnet infections and Botnet C&Cs

Please no more "technical solutions" on members-discuss!

- Cynthia

On Thu, Apr 30, 2020 at 10:31 PM Elad Cohen <elad at netstyle.io<mailto:elad at netstyle.io>> wrote:
Hello Ripe Members!

I will share the following solution in the near General Meeting and I'm interested to share the following technical solution with you as well, it will completely resolve: Spoofed IP traffic, Spoofed amplification DDoS attacks, BGP&RIR hijacking. And will dramatically lower: IoT botnet infections and Botnet C&Cs.

By a single update to any BGP router, not any router needs to be updated, only active BGP routers. If I will have the honor of being elected to the Ripe Board I will harness all the power of Ripe and all of the 5 RIR's and all of the LIR's in the 5 RIR's so routing manufacturing companies will implement the below processes and methods with a single firmware update to their BGP routers.

I'm asking for your support in electing me so I will be able to enter the Ripe Board and then I will be able to make everything which is written in this post to come true.


Regarding the bgp-anycasted infrastructure written below, ICANN is written but the global bgp-anycasted infrastructure can be managed by the 5 RIR's and/or by the ccTLDs registries (my main point is that who will operate the bgp-anycasted infrastructure is not important for now, as long as it will be an agreed authoritative non-governmental non-commercial global entity/ies)

With new tracking protocol over ip, routers will be able to confirm that source ip came from the network of the announcing ASN, and hence spoofed amplification DDoS attacks will be completely annihilated.


The Process:

At the source BGP router, for any ip packet with a source address that is from the network of the source BGP router (lets call it original ip packet) - the source BGP router will create a new ip packet (lets call it tracking ip packet) with a new transport layer protocol and with the same source address and with the same destination address and with the same IP-ID such as the original ip packet.

In the new tracking ip packet there will be a new transport layer protocol (tracking protocol) with the following fields:
AS number of source BGP router in clear text
AS number of source BGP router encrypted with the private key of the source BGP router

The destination BGP router (a BGP router that the destination address is in its network) whenever it receive a 'tracking ip packet' will check if its have the internal boolean 'Check tracking flag' in it - 'on' or 'off':
If 'off' then the destination BGP router will drop that 'tracking ip packet'

If 'on' then the destination BGP router will decrypt the 'encrypted AS number' with the public key of the specific AS number
and after decryption the AS number need to be the result:
if not then to drop the tracking ip packet and the original ip packet related to it
if yes then to drop the tracking ip packet and to forward the related original ip packet to destination but only if the source address is originated from the specific ASN (according to the local ASNs+ranges table in the BGP router, such table will be received from ICANN)


If the 'Check tracking flag' is set to 'on' then any original ip packet that arrive to the destination BGP router will wait for the related tracking ip packet (in case the related tracking ip packet didn't already arrived to the destination BGP router). The destination BGP router will manage such waiting for X number of seconds.

The destination BGP router will match between a tracking ip packet and an original ip packet - based on their source address and their destination address and their IP-ID which will all be identical.



More Aspects:

- The end-devices will not need to be updated, any router will not need to be updated, only all the BGP routers will need to be updated.
- Any BGP router in the routing path, which the original ip packet and the tracking ip packet are not destined to an ip address in its own network - will not check the content of the tracking ip packet and will forward both the tracking ip packet and the original ip packet as they are.
- Each BGP router will have all the public keys (of all the ASN's) locally.
- Each BGP router will have a full list of all the ASN's and all the route objects ranges which are related to them locally.



How BGP routers will receive all the ranges in all the route objects of all the ASNs (in the 5 RIRs) and all the public keys of all the ASNs (for decrypting the encrypted strings in 'tracking ip packets'):

- Each BGP router will create a tcp session with ICANN backend infrastructure (the backend infrastructure of ICANN will use BGP anycast and will be available from many locations worldwide with automatic syncing)
- At this stage there will be a handshake process between the BGP router and the ICANN backend infrastructure in order for ICANN to know the correct ASN which is operating the BGP router - the BGP router will send its ASN in cleartext and also its ASN encrypted with its ICANN-communication-private-key , ICANN will know the related public key for the specific ASN from the specific ASN object in the RIR (the public key for communication with ICANN will be displayed there) - after decryption ICANN will compare the decrypted string to the AS Number for successful authentication.
- After successful authentication, all the communication will be encrypted, ICANN will notify the BGP router about its public key and ICANN will use the public key of the ASN for the communication with ICANN - from the ASN object in the RIR.
- The BGP router will send over the session its public key to be used by other BGP routers in order to decrypt the encrypted string in the tracking ip packets that it will originate (that private key and public key will be managed in the BGP router GUI or CLI).
- ICANN will notify all the other BGP routers through the sessions with them about a newly updated such public key of any other BGP router.
- ICANN will also receive in real-time any route object creation/modification/deletion notification from any of the 5 RIRs and will then update all the BGP routers through all of their sessions.

- In case a BGP router doesn't have an active session to ICANN backend infrastructure (for any reason, might be due to networking issue) - then temporarily the internal 'Check tracking flag' of it will be set to 'off'. After the session with ICANN backend infrastructure will be re-established and the BGP router will receive all the meantime updates - the boolean value of 'Check internal flag' will return to initial state.
- Any update from ICANN backend infrastructure to a BGP router (such as a public key of another BGP router, or a routing object update) - will be confirmed that the update was received well by the BGP router side.



'Check tracking flag' in BGP Routers:

- BGP routers, in their GUI and CLI interfaces - will not allow the end-user to set the boolean value of 'Check tracking flag', in order to avoid misconfiguration.
- The ICANN backend infrastructure through the session with the BGP router - will be able to set the boolean value of the 'Check tracking flag'.
- The reason for it, is that if 'Check tracking flag' will be set on some destination BGP routers while some other source BGP routers weren't upgraded yet (and will not send the 'tracking ip packet' due to it) - then 'tracking ip packet' will never reach the destination BGP router and the internet will break.
- Central setting of 'Check tracking flag' through ICANN backend infrastructure - will allow ICANN to inform all the BGP routers at once to switch 'on' the 'Check tracking flag'
- ICANN, in the session to any BGP router, will also receive the percentage of ip packets that were destained to that BGP router network - that also had ip tracking packets, in this way ICANN will know when all the BGP routers were properly globally updated and when it is time to enable the 'Check tracking flag' in all the BGP routers.
- ICANN will know if all the BGP routers in the world were upgraded based on keeping the full BGP table and comparing it to all the BGP routers that did and that did not open a session to ICANN backend infrastructure.



Automatic preventation of IoT botnet infections:

- IoT botnets are based on default credentials, if we can block default credentials of IoT devices then these kind of botnets (such as Mirai-variants and similar) will stop to have an impact in the internet.
- The data field in an ip packet - will always be the same for an access attempt to a IoT device with default credentials - hence these kind of "IP protocol data fingerprints" which are related to specific "IP protocol numbers" will be provided by ICANN backend infrastructure to each BGP router through the opened session with it.
- There are two issues with matching incoming ip packets to the "locally stored IP protocol data fingerprints" - the first one is that ip packets can be sent by fragments (so not all the data field will be sent at once in order to be able to be compared with the locally stored data fingerprints) and the second is that usernames (or url's) or any other textual data in the incoming ip packet data field can be in uppercase or in lowercase. In order to overcome the possibility of the existence of a single data fingerprint in multiple incoming ip packet fragments - then in case the BGP router is recognizing the incoming fragmented ip packet data value as part of an existing fingerprint data in its local database then it will keep track of the arrival ip packet fragments based on their specific IP-ID identifier and the BGP router will not forward the last ip packet fragment if the last ip packet fragment will cause all the related ip packet fragments to match a specific ip fingerprint data (last ip packet doesn't have to be the last fragmented part but it is the last ip packet that arrived with that IP-ID identifier, so the BGP router will keep track of the specific fragmented IP packets that arrived and their indexes in order to know when the last one of them arrived). Regarding the second issue - the stored data fingerprints in the local BGP router will be stored in a way that some bytes of them (in specific indexes) will not be compared and in case all the other bytes will match - then the bytes in these indexes - will first be lowered case - and only then comparison will be made to the specific indexed bytes in the specific ip packet data fingerprint.
- In case a IoT device behind a BGP router will be infected somehow (for example when a specific fingerprint data with default credentials for a specific device wasn't updated yet through ICANN backend infrastructure), it will be able to infect all the other IoT devices in the local network when the connectivity to them is not through the BGP router, that kind of impact will be much much lower than infected IoT device which can infect any other IoT device in the internet and then massive botnets in the internet are created which are being used for DDoS.



Automatic prevention of botnet C&C ip addresses:

- Botnets C&C are also a problem in the internet.
- This problem can be overcome using the following technical addition: the 5 RIR's will operate end-users honeypots machines all over the world (it will be implemented by a single physical machine in each location, for example in each datacenter and in each major ISP, each single physical machine will emulate a virtual router and virtual VM's, the virtual VM's will emulate many different kinds of 'real world machines', any kind of automatic updating (in the operating system configurations) will be disabled, these honeypots machines are not intended to make any outgoing connection, the virtual routers will monitor if any outgoing connection is made and if yes then it is to a botnet C&C, the virtual router will update the ICANN backend infrastructure regarding it and the ICANN backend infrastructure will update all the BGP routers (in their open sessions) regarding it to completely block any communication to that botnet C&C ip address. There will be a web-based system and only the related Law Enforcement Agency of that C&C ip address region - will be able to remove that C&C ip address from being blocked after their manual check.
- Honeypot machines will be deployed using 'templates' - these templates must be signed and not anyone can create them, they should be created and signed by an agreed Law Enforcement Agency such as Interpol in order to make sure that these templates are by-design not making any outgoing connection. The templates will be deployed in an automatic way (major ISP's and datacenters will be able to easily add a 'physical honeypot' server in their network, that will be shipped to them), the re-initiation of a compromised 'virtual machine' that made  an outgoing connection - will also be automatic through the system in the physical server.


Respectfully,
Elad

_______________________________________________
members-discuss mailing list
members-discuss at ripe.net<mailto:members-discuss at ripe.net>
https://mailman.ripe.net/
Unsubscribe: https://lists.ripe.net/mailman/options/members-discuss/me%40cynthia.re
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://www.ripe.net/ripe/mail/archives/members-discuss/attachments/20200430/bd180e18/attachment.html>