This Is Probably Why Half the Internet Shut Down Today


#21

I’m too always thinking with this possibility, and the solution must come that the churn of recent nodes have minimal impact in the group.
The union of Disjoint Group and Node Ageing go to the good direction and can make that the churn of nodes with low age will not cause too much problems to the network.
The problem would be more serious if someone have enough computational power to control the 1% of high age nodes. In this case the impact to the network can be much more harmful.

And, as you note, a massive Group Merging could be the nightmare of the Safe network.


#22

With the combination of CJDNS, this is impossible.


#23

I got an idea when I saw the picture in the article of the mess of blue cables. The image looked similar to other messy and cumbersome technologies in the past such as manual telephone switchboards.

Here is an even worse example of network cables:

And here is a picture of the wires inside an old supercomputer:

It made me think that we will probably have a new “tidier” technology that will replace today’s network cables. And that TCP/IP may be replaced by another form of network protocol.


#24

TCP/IP was actually for a network based on a loop of wire. So ONE wire (coax) went from one port to the next.Thus one loop would connect a whole rack.

The reason for the mess of wires is simply SPEED. Switches allow the reduction of traffic by segmenting the traffic.

Its more to do with the underlying layers than the protocol.


#25

Many people have speculated about a “killer” app that will launch Safe Network into the mainstream. Look no further than this: Whenever necessary or important services cease to function, the inconvenience of establishing new connections (that is, to SafeNet) will be secondary to the urgency of maintaining stability. Profit motives and customer outrage will catapult major sites into the realm. After a few have tested the waters, there will be a headlong rush to “sign up”. These DDOS assaults being carried out via the hijacking of thousands or millions of IoT devices will insure the adoption of SafeNet much more quickly than most anticipate. Get ready.


#26

Two solution are on my mind:

  1. Let the nodes be “snobs” and ask proof of work, depending on their load. In fact, I think on some cryptographic ways that the nodes who host the data could prove their load is high by showing old request proof of work(but that requires timestamping, but again we can pass this last suggestion). Another way to discourage nodes from being “snobs” is by make them sacrifice their farming rate(a balance between better trafic and less proof of work).
  2. Similarly to what said in previous posts, nodes could ask for emergency backup from their neighborhoods. However, that leads to the question how to reach consensus about the version of the data. I’m not good in statistics, but does multiple verification from random-close-enogh nodes do the trick?

Notice that this high load problem doesn’t happen only on DDoS attacks. For example, maybe everyone just want to see a crazy popular video in SAFE? OK, you’ll say it’s an immutable data and everyone can help with no verification problem… But lets assume it’s a structure data for the example.


#27

“If you do a Google patent search you will see the state-of-the-art there,” Maidsafe CEO David Irvine told Hacked.com. “It’s very new.”

Maidsafe, and it’s SAFE network is modelled on colonies in nature, servers and intermediaries do not make logical sense and we sort of all know that in our deepest thoughts. Some say a decentralized internet is not possible.

“DDOS depends on a target, in SAFE the target is everyone’s computers,” Mr. Irvine said. “An analogy would be DDOS is like swatting a large fly, in SAFE the large fly is a mass swarm of mossies which makes it hard even with 8 arms filled with fly swatters.” Mr. Irvine hopes the SAFE network might help to form the primordial soup of a decentralized internet.

SAFE uses xor addressing, a type of data structure used in computer programming, which means any node outages will be covered by the nearest nodes.

“This makes those mossies like zombie mossies or perhaps hydra like,” Mr. Irvine said.

This is stretching a little, he admits. Since no such decentralized networks exist, it’s difficult to know what one’s vulnerabilities might be. Maidsafe employs university research to help answer some questions. Maidsafe uses data chains, which are different from blockchains.

The Maidsafe model does not assume every node/person needs to know and agree on all information like in the Bitcoin blockchain model. “Detailed information is local and less detailed is more global,” he said. “This is like posting a letter, the closer to the destination the more info is known of the destination.”

https://hacked.com/decentralized-internet-solution-outages/


#28

Do you mean specifically the CJDNS project, or the ideas it contains as implemented by the safe network protocol?

Thanks for mentioning it, I hadn’t heard of this before. It’s a very interesting project.


#29

That’s only for TCP-based DDOS attacks though. Attackers have lots of other options. For example, they could just saturate your connection with UDP or ICMP packets, right? No need to establish a connection of any kind, and by the time the packets reach your computer the damage is already done; your downlink connection has been slowed to a crawl by a flood of useless traffic.

Unless MaidSafe is somehow able to hide the IP addresses of its nodes, I don’t think there’s any way to prevent individual nodes from getting taken out by DDOS. The real advantage of MaidSafe against DDOS attacks is that individual nodes going down shouldn’t significantly impact the network as a whole, right?


#30

Correct. The meshnet project.

Safenet runs on top of the CJDNS protocol. Have couple trusted buddies in your rural area connect to your node, and visa versa. A connects to B, B connects to C, A can communicate to C without knowing C IP address. The coolest part about this is that Safenet requires a person to connect to the routing table before it given XOR address. Your trusted friend(cjdns node (safenet routing table)) will provide you XOR address. Not some random stranger on monopolistic ISP system. The first hop is the most critical part of the whole security and trust. After the first hop, nobody knows where it came from and where it’s going.

Impossible. How does one communicate without knowing IP address? That’s like saying, I am a tree, who wants to communicate to another tree without having the roots.


#31

Tor allows this. As does I2P. AFAIK neither of those will protect against DDOS of public servers though, which was kinda my point. This is a very difficult problem to solve, and MaidSafe would be (and indeed is) better served by an alternate strategy for DDOS protection, such as being able to tolerate the loss of individual nodes.


#32

Tor and I2P still uses IP address. It doesn’t mask your IP address at all. The only thing that protect your identity, and anonymity is the onion/garlic routing. The ISP knows you’re connected to TOR/I2P network but they don’t know who you’re communicating to.


#33

Of course your ISP knows your IP address. You’re not trying to protect against DDOS attacks from your ISP though. It’s everyone else you have to worry about, and those are the people TOR and I2P prevent from learning your IP Address.


#34

This new RFC, from Diggory Hardy, tries to solve the problem of cascade group merging by the creation of subgroups. With the existence of this subgroups the merging nodes are no longer required to grab all data, matching the group they merge, reducing network load.


#35

Not trying to be a downer but… how many more problems need to be resolved. :disappointed:
All the tutorials are grand but really hoping for some inspiring light in the network tunnel.


#36

“Rome wasn’t built in a day”