Proof of work comes to Tor | Bitcointalksearch.org

ABCbits

legendary

Activity: 2870

Merit: 7490

Crypto Swap Exchange

Quote from: NotATether on June 06, 2023, 05:38:46 AM

Quote from: ?? on ??

--snip--

In the more extreme case, a TOR relay's proxy server could theoretically cache this information on a CDN such as Akamai, which would then transform the problem into a DDoS attack on said CDNs, which could cause them to sinkhole the IP addresses involved. Or perhaps the sinkholing can be done by the relays themselves with the proper network infrastructure, since I don't know if it's even practical for CDNs to get involved, even if it is in the very early stages before the bridges/guard nodes do anything else.

I really hope people who run Tor hidden service wouldn't resort to 3rd party CDN. Even OnionBalance and Endgame which created for Tor hidden service isn't recommended by Tor project (as stated by article shared by OP).

Quote from: darkv0rt3x on June 06, 2023, 09:56:31 AM

An out of the blue question:
I2P seems to be using hashcash as means of preventing email spamming and DDoS (back in the days by Adam Back, iirc)... I2P is way newer than Tor so why is Tor only now adopting a more effective way of dealing with DDoS?

I don't know since when I2P add Hashcash. But PoW proposal for Tor has been around since 2020[1] and the idea itself seems to be few years older. And FYI, I2P isn't exactly new since it was created on 2003[2].

Quote from: darkv0rt3x on June 06, 2023, 09:56:31 AM

They did use anything else before now?

Yes, some of them mentioned on their documentation[3].

[1] https://lists.torproject.org/pipermail/tor-dev/2020-June/014381.html
[2] https://www.geti2p.net/en/blog/post/2021/08/28/History-of-I2P
[3] https://community.torproject.org/onion-services/advanced/dos/

darkv0rt3x

hero member

Activity: 1232

Merit: 666

I rather die on my feet than to live on my knees

An out of the blue question:
I2P seems to be using hashcash as means of preventing email spamming and DDoS (back in the days by Adam Back, iirc)... I2P is way newer than Tor so why is Tor only now adopting a more effective way of dealing with DDoS? They did use anything else before now?

NotATether

legendary

Activity: 1568

Merit: 6660

bitcoincleanup.com / bitmixlist.org

Quote from: ?? on ??

Quote from: larry_vw_1955 on June 05, 2023, 11:19:41 PM

Quote from: ?? on ??

Providing PoW parameters (on step 1) is extremely cheap since the parameter is only updated every ~2 hours. See section 3.1 for details. And IMO few μs to verify PoW is very fast.

i was talking about step 2 regarding network bandwidth:
2) Client fetches descriptor and computes PoW [CLIENT_POW]

someone could still try and saturate the server's bandwidth/ability to process requests for the "descryptor" a better solution would not require the server to send out anything to the client until after the client successfully submitted a valid PoW...

But server need to send description which contain PoW parameter in order for client to perform and submit valid PoW in first place. And AFAIK sending descriptor to client should be extremely cheap since descriptor is static data which only updated as needed (e.g. every ~2 hours to update PoW parameters) and could be cached easily.

In the more extreme case, a TOR relay's proxy server could theoretically cache this information on a CDN such as Akamai, which would then transform the problem into a DDoS attack on said CDNs, which could cause them to sinkhole the IP addresses involved. Or perhaps the sinkholing can be done by the relays themselves with the proper network infrastructure, since I don't know if it's even practical for CDNs to get involved, even if it is in the very early stages before the bridges/guard nodes do anything else.

larry_vw_1955

sr. member

Activity: 1036

Merit: 350

Quote from: ?? on ??

Providing PoW parameters (on step 1) is extremely cheap since the parameter is only updated every ~2 hours. See section 3.1 for details. And IMO few μs to verify PoW is very fast.

i was talking about step 2 regarding network bandwidth:
2) Client fetches descriptor and computes PoW [CLIENT_POW]

someone could still try and saturate the server's bandwidth/ability to process requests for the "descryptor" a better solution would not require the server to send out anything to the client until after the client successfully submitted a valid PoW...

Quote

[2] https://thecyberthrone.in/2020/08/21/tor-finally-fixed-a-bug-that-annoyed-of-ddos/

“The attacks exploit the inherent asymmetric nature of the onion service rendezvous protocol, and that makes it a hard problem to defend against. During the rendezvous protocol, an evil client can send a small message to the service while the service has to do lots of expensive work to react to it. This asymmetry opens the protocol to DoS attacks, and the anonymous nature of our network makes it extremely challenging to filter the good clients from the bad.”

thats a poorly designed protocol in the first place then.

larry_vw_1955

sr. member

Activity: 1036

Merit: 350

Quote from: tromp on June 04, 2023, 03:43:50 AM

You're right that a DDOS attacker can submit bogus PoW proofs at negligible cost and these still need to be verified. The HashX docs make these claims regarding Performance [1]

Thanks for the info but:

A few comments on this proposal: https://github.com/torproject/torspec/blob/main/proposals/327-pow-over-intro.txt

Quote from: Cricktor on June 04, 2023, 05:46:34 PM

I'm not a Tor expert either. I read that those Tor rendezvous circuits are something "expensive" for the Tor participants, whatever that means. Apparently it made sense to implement this PoW challenge for those rendezvous circuits, if needed under some DDoS stress. Usually it's computationally way more expensive to do the PoW than to check if correct PoW has been done and submitted. It's likely similar to what is done in Bitcoin mining. Finding a valid block hash is multi-magnitudes of computational work more expensive than validating the correctness of a Bitcoin block header hash.

strange thing is how they can turn off the PoW when it's not needed. But clients can still do the PoW anyway to get a higher priority in the "Queue". So people with more powerful machines (modern ones), well they can be expedited while poorer people with lower end machines can get ignored. Ignored means treated with lower priority. means more waiting. Shocked

Cricktor

hero member

Activity: 714

Merit: 1010

Crypto Swap Exchange

Quote from: larry_vw_1955 on June 03, 2023, 11:38:44 PM

but they still have to take requests and check if the appropriate amount of work has been done. that takes resources just doing that.

I'm not a Tor expert either. I read that those Tor rendezvous circuits are something "expensive" for the Tor participants, whatever that means. Apparently it made sense to implement this PoW challenge for those rendezvous circuits, if needed under some DDoS stress. Usually it's computationally way more expensive to do the PoW than to check if correct PoW has been done and submitted. It's likely similar to what is done in Bitcoin mining. Finding a valid block hash is multi-magnitudes of computational work more expensive than validating the correctness of a Bitcoin block header hash.

o_e_l_e_o

legendary

Activity: 2268

Merit: 18587

There is a good layout of the various attack surfaces for this approach, and how they can be mitigated against, starting here: https://github.com/torproject/torspec/blob/main/proposals/327-pow-over-intro.txt#L631

tromp

legendary

Activity: 984

Merit: 1091

Quote from: larry_vw_1955 on June 03, 2023, 11:38:44 PM

but they still have to take requests and check if the appropriate amount of work has been done. that takes resources just doing that.

You're right that a DDOS attacker can submit bogus PoW proofs at negligible cost and these still need to be verified. The HashX docs make these claims regarding Performance [1]

> HashX was designed for fast verification. Generating a hash function from seed takes about 50 μs and a 64-bit nonce can be hashed in under 100 ns (in compiled mode) or in about 1-2 μs (in interpreted mode).

Obviously, with hash function generation being about 3 orders of magnitude slower than a SHA256 hash computation, fast verification was not the prime design objective.

It was designed for ASIC resistance first, and fast verification second.

The EquiX PoW verification makes 8 HashX computations, but they're all from a single generated function, so detecting the bogus PoW takes about 50 μs. Hopefully that's fast enough to deter such attacks.

EDIT: apparently TOR already faces about 260 μs overhead in request processing before it can verify the PoW, so there's very little benefit in further reducing this 50 μs.

[1] https://github.com/tevador/hashx

larry_vw_1955

sr. member

Activity: 1036

Merit: 350

Quote from: o_e_l_e_o on June 03, 2023, 07:53:17 AM

So it's not an all or nothing thing as I had initially assumed - the service stipulates x amount of work, if you complete x or more you are connected, if you complete less than x you are not connected. Rather, it's based on prioritization. If you complete less than the suggested amount then you might still connect, but your connection will simply be a lower priority than those who have completed more work.

but they still have to take requests and check if the appropriate amount of work has been done. that takes resources just doing that. you could overload a server with that type of bogus request and what are they going to do? Shocked

i'm not an expert on tor.

Synchronice

hero member

Activity: 882

Merit: 792

Watch Bitcoin Documentary - https://t.ly/v0Nim

Quote from: o_e_l_e_o on June 02, 2023, 07:18:05 AM

There's a fantastic post here which explains how this is going to work and what the end user will experience: https://darkdot.com/articles/tor-ddos-leads-to-proof-of-work/. Site admins can manually decide how much PoW is required to reach their site, and users can decide how much PoW they are willing to perform.

Hopefully this puts an end to DDoS on Tor.

I think you slightly misunderstood that part. Service administrator can manually enable or disable PoW for an Onion service, i.e. the administrator can leave it disabled if there is not an attack or enable it in case there is an attack. Also, website administrator can't manually decide how much PoW is required for individuals to reach their site, it's automatically adjusted by the tor Daemon, based on the scope of the attack.

Wind_FURY

legendary

Activity: 2898

Merit: 1823

Quote from: NotATether on June 02, 2023, 01:43:18 PM

Quote from: o_e_l_e_o on June 02, 2023, 07:18:05 AM

Hopefully this puts an end to DDoS on Tor.

Oh it will alright, but it will also get Greenpeace's angry wrath Cheesy

At this point, these watchdogs should just go after Netflix, your IDE, software updater, and all your games for your COU contributing to the carbon footprint. We shall hear what they have to say about this eventually, if ~~their XRP fallguy~~ they even care at all Roll Eyes

Don't worry about "Green Peace", their people at the top don't actually care. Tin-foil hats on, but just like the World Economic Forum, it's merely one of the organizations created to help achieve the elite's Globalist Agenda.

Quote from: She shining on June 02, 2023, 08:59:24 AM

Saw online while trying to understand more on the attack a term called hactivist

Quote

https://gridinsoft.com/blogs/ddos-for-hire-used-by-hacktivists/ The reason for the recent uprising of DDoS attack and, particularly, DDoS-for-hire services, is hacktivists activity. Hacktivism has evolved from loosely structured groups to a more mature ecosystem with diverse motivations and sources. It got a massive punch particularly after the beginning of the Russia-Ukrainian war.

As a result, hacktivist groups have become more organized and conduct military-like operations with precise positioning and clear objectives

won't this organized attack pose a threat to the system in its early adoption? But in all I wish the project success because having a network that boast of good anonymity is uncommon

It will, just like how miners could be hired and be a threat against nascent altcoins, BUT it's not without a cost which is the point of POW. Want to attack it? Pay the price because sometimes it won't be cheap.

Cricktor

hero member

Activity: 714

Merit: 1010

Crypto Swap Exchange

I understood it that this PoW demand for a circuit rendezvous is only issued when the Tor service detected some kind of DDoS situation. In a network or connection pressure situation the PoW will be enforced to mitigate the pressure otherwise it's not enabled and demanded by the Tor service. I like this approach and the idea behind it.

Will be interesting to see if it works well and does what it's designed for.

PrimeNumber7

copper member

Activity: 1624

Merit: 1899

Amazon Prime Member #7

Quote from: o_e_l_e_o on June 03, 2023, 07:53:17 AM

I'm really keen to try this out. Wonder how long it will be before some sites start implementing it?

I assume this is only an option for hidden services (so only .onion sites).

Although facebook is not very privacy friendly to most (all?) of their users (based on the amount of data they collect on them), they do have a tor version of their site, and I can see them implementing this as a means to help others to implement this to prevent DoS attacks on hidden services.

o_e_l_e_o

legendary

Activity: 2268

Merit: 18587

Something interesting I read when looking more closely at this:

Quote from: https://gitlab.torproject.org/tpo/core/tor/-/blob/8b46d1c6ca20b8c99b979569c7432a97d8fc20a1/doc/man/tor.1.txt#L3102-3106

Incoming rendezvous requests will be prioritized based on the amount of effort a client chooses to make when computing a solution to the puzzle. The service will periodically update a suggested amount of effort, based on attack load, and disable the puzzle entirely when the service is not overloaded.

So it's not an all or nothing thing as I had initially assumed - the service stipulates x amount of work, if you complete x or more you are connected, if you complete less than x you are not connected. Rather, it's based on prioritization. If you complete less than the suggested amount then you might still connect, but your connection will simply be a lower priority than those who have completed more work.

I'm really keen to try this out. Wonder how long it will be before some sites start implementing it?

gmaxwell

staff

Activity: 4172

Merit: 8419

Quote from: tromp on June 02, 2023, 09:31:13 AM

Quote from: She shining on June 02, 2023, 08:59:24 AM

Quote

https://darkdot.com/articles/tor-ddos-leads-to-proof-of-work/ it was actually invented as an anti-spam measure by Adam Back in 1997

I thought POW was created by Moni naor and Cynthia Dwork.

Naor & Dwork invented the concept of PoW, while Adam invented Hashcash, the first practical PoW algorithm [1].

[1] https://en.wikipedia.org/wiki/Proof_of_work

Context: scammer Craig Wright pays people to spread that line because: (1) He has some narcissistic rivalry with everyone who understands technology better than him (which is lot of people!), (2) he doesn't understand proof of work well enough to realize that a version with a trap-door is fundamentally unsuitable for almost every application people talk about using proof of work. It's just pathetic.

Unfortunately, the whole thing about history being written by the victors only applies to stuff like warfare where the losers are literally killed. Lots of people who don't know better end up picking up and repeating his narratives-- he stands to gain from spending to spread his bullshit and invests heavily in it. Other people don't care to stop him. Thanks for speaking up for the truth.

NotATether

legendary

Activity: 1568

Merit: 6660

bitcoincleanup.com / bitmixlist.org

Quote from: o_e_l_e_o on June 02, 2023, 07:18:05 AM

Hopefully this puts an end to DDoS on Tor.

Oh it will alright, but it will also get Greenpeace's angry wrath Cheesy

At this point, these watchdogs should just go after Netflix, your IDE, software updater, and all your games for your COU contributing to the carbon footprint. We shall hear what they have to say about this eventually, if ~~their XRP fallguy~~ they even care at all Roll Eyes

Quote from: She shining on June 02, 2023, 08:59:24 AM

Saw online while trying to understand more on the attack a term called hactivist

Quote

.https://gridinsoft.com/blogs/ddos-for-hire-used-by-hacktivists/ The reason for the recent uprising of DDoS attack and, particularly, DDoS-for-hire services, is hacktivists activity. Hacktivism has evolved from loosely structured groups to a more mature ecosystem with diverse motivations and sources. It got a massive punch particularly after the beginning of the Russia-Ukrainian war. As a result, hacktivist groups have become more organized and conduct military-like operations with precise positioning and clear objectives

won't this organized attack pose a threat to the system in its early adoption? But in all I wish the project success because having a network that boast of good anonymity is uncommon

The internet is not a WWII battlefield and should be treated as neutral protected space. This especially goes for IP addresses running important and constructive protocols like DNS and Tor. Hacktivists should not be allowed to abuse the system as well.

tromp

legendary

Activity: 984

Merit: 1091

Quote from: She shining on June 02, 2023, 08:59:24 AM

Quote

https://darkdot.com/articles/tor-ddos-leads-to-proof-of-work/ it was actually invented as an anti-spam measure by Adam Back in 1997

I thought POW was created by Moni naor and Cynthia Dwork.

Naor & Dwork invented the concept of PoW, while Adam invented Hashcash, the first practical PoW algorithm [1].

[1] https://en.wikipedia.org/wiki/Proof_of_work

She shining

jr. member

Activity: 38

Merit: 24

Quote

https://darkdot.com/articles/tor-ddos-leads-to-proof-of-work/ it was actually invented as an anti-spam measure by Adam Back in 1997

I thought POW was created by Moni naor and Cynthia Dwork.

Tor have been on this project for a long time now,i know its difficult to implement but still think their adoption was kinda slow and passive. According to the link the project is implementing a form of proof of work similar to POS,but instead of stakes works are staked,which can really help tackling small,mid or maybe large botnet but won't this make the system way slower? Can a computer with relatively poor computation power still effectively use it?

Saw online while trying to understand more on the attack a term called hactivist

Quote

.https://gridinsoft.com/blogs/ddos-for-hire-used-by-hacktivists/ The reason for the recent uprising of DDoS attack and, particularly, DDoS-for-hire services, is hacktivists activity. Hacktivism has evolved from loosely structured groups to a more mature ecosystem with diverse motivations and sources. It got a massive punch particularly after the beginning of the Russia-Ukrainian war. As a result, hacktivist groups have become more organized and conduct military-like operations with precise positioning and clear objectives

won't this organized attack pose a threat to the system in its early adoption? But in all I wish the project success because having a network that boast of good anonymity is uncommon

tromp

legendary

Activity: 984

Merit: 1091

Quote from: o_e_l_e_o on June 02, 2023, 07:18:05 AM

It uses the Equi-X algorithm, which is itself developed from RandomX, which is the ASIC resistant PoW algorithm used by Monero.

To be more precise, we should distinguish between PoW algorithms and the hash functions they use.

E.g. when people say Bitcoin's PoW is SHA256, what they mean is that it uses the Hashcash PoW with the SHA256 hash function (iterated twice).

Equi-X is the Equihash PoW algorithm using HashX as underlying hash function.
The latter is derived from the RandomX hash function used in Monero's Hashcash PoW algorithm.

o_e_l_e_o

legendary

Activity: 2268

Merit: 18587

https://gitlab.torproject.org/tpo/core/tor/-/commit/8b46d1c6ca20b8c99b979569c7432a97d8fc20a1

Quote

o Major features (onion service, proof-of-work):
   - Implement proposal 327 (Proof-Of-Work). This is aimed at thwarting
   introduction flooding DoS attacks by introducing a dynamic Proof-Of-Work
   protocol that occurs over introduction circuits. This introduces several
   torrc options prefixed with "HiddenServicePoW" in order to control this
   feature. By default, this is disabled. Closes ticket 40634.

It uses the Equi-X algorithm, which is itself developed from RandomX, which is the ASIC resistant PoW algorithm used by Monero. It is also being developed by the same user, tevador.

There's a fantastic post here which explains how this is going to work and what the end user will experience: https://darkdot.com/articles/tor-ddos-leads-to-proof-of-work/. Site admins can manually decide how much PoW is required to reach their site, and users can decide how much PoW they are willing to perform.

Hopefully this puts an end to DDoS on Tor.

Topic: Proof of work comes to Tor (Read 366 times)