Should Linus do a fast as possible on IPv6?

[faftek]

With the recent need for IPv6 and it being implemented in a lot of areas, we should really get an idea of how IPv6 works and compare it to IPv4. Everyone please vote, i'm begging you. IF YOU CLICK YES THE NSA WON'T SPY ON YOU.

 

Oh and this: http://t.co/52zw5kXJi7

 

ヽ༼ຈل͜ຈ༽ﾉ raise your dongers for IPv6 ヽ༼ຈل͜ຈ༽ﾉ

 

EDIT: Wow it seems like a heck of a lot of people want to hear about IPv6

 

If Linus actually considers doing this I'll write him a full walkthrough of everything to know about IPv6. 

[YeroJero]

All the yes

[Jodomcfrodo]

Considering I have no fracking idea what IPv6 is, Imma say yes. 

[Poohy14]

I think it should be the difference between IPV6 and IPV4

[Flight1sim]

Why not?

[Mokedoke]

I agree!

[cj09beira]

yes. i have no idea how that works and i dont have time to read something about it am lazy

[xmx1024]

I kinda understand it, but it would be nice to fully understand it.

 

yes.

[Stefan Becker]

I don't know what the hell that is so... Yes?

[Ryan.L]

yes

[Dragoneye3769]

THAT 1 GUY

[Liam-McShane]

Why? No one uses it.

[Paralectic]

Why? No one uses it.

Not yet, in several years we will use it.

[TheWall]

@LinusTech i don't watch a lot of techquikie vids because i already understand most of them fairly well or i don't care about them but i would really love this explained to me simply

[Me1z]

Why? No one uses it.

 

It has not been implemented everywhere yet, but most ISPs in the US are currently running both v4 and v6 right now and you are assigned one of each. Once there is enough support for it v4 will be phased out and only used for cases where legacy hardware does not support ipv6. It's not like switching from v4 to v6 is optional or a preference, we literally ran out of ip addresses back in 2011. There is currently a waiting list set up to distribute address blocks as they become available, but that will only work for so long. There are even smaller ISPs that are coping with this by only assigning addresses to devices that are actively in use and reassign the address after a certain period of inactivity. While somewhat functional, this rather sucks because static ip addresses are much nicer to work with and allow you to host private servers without paying 'enterprise level' prices. The sooner we kill off IPv4 and implement v6 the better IMHO.

[Madgemade]

Not much point as it's pretty technical and consumers won't be very affected.

BUT STILL FUN TO DO!

 

also in the UK only one consumer provider gives ipv6 and very few people use them so not widespread here.

yes for hosting ipv6 is a dream come true as here we all have dynamic ipv4 to get static must pay around extra $20(£15?) a month with only a few providers who are already pricey

[LtRoyalShrimp]

Maybe also add a tiny explanation of why the Xbox One sucks in its implementation of IPv6. 

 

To get it via Apple routers: Go into the airport settings, find IPv6, enable tunneling. Now your Xbox One will be able to use IPv6 as well as no more multiplayer issues(unless your NAT is strict). 

 

All other routers: Find your IPv6 settings, look for something like "IPv6 Tunneling" and enable it. 

 

It was stupid of Microsoft to enable this workaround when IPv6 is readily available these days. 

[faftek]

Wow it seems like a heck of a lot of people want to hear about IPv6

 

If Linus actually considers doing this I'll write him a full walkthrough of everything to know about IPv6. 

[LAwLz]

Not much point as it's pretty technical and consumers won't be very affected.

BUT STILL FUN TO DO!

 

also in the UK only one consumer provider gives ipv6 and very few people use them so not widespread here.

yes for hosting ipv6 is a dream come true as here we all have dynamic ipv4 to get static must pay around extra $20(£15?) a month with only a few providers who are already pricey

I just use dynamic DNS instead. Far easier than getting a static IP.

[Hikaru12]

I just use dynamic DNS instead. Far easier than getting a static IP.

 

Cheaper too.

[Jaybird]

IPv6 is nearly exactly like IPv4. the only major difference is it uses 128 bit address instead of 32 bit. thus allowing for more address then IPv4. after that there nothing really interesting to talk about. 

[The_Strict_Nein]

Donger raised

[faftek]

IPv6 is nearly exactly like IPv4. the only major difference is it uses 128 bit address instead of 32 bit. thus allowing for more address then IPv4. after that there nothing really interesting to talk about. 

At first glance it seems that way, but there's actually a lot more to it, as well as security issues that can arise out of it. For example each computer now gets it's own IP to the internet, it's not one IP to the location(or however many you pay for) and then the devices share the external IP and only get an internal one.

[juretrn]

IPv6 is nearly exactly like IPv4. the only major difference is it uses 128 bit address instead of 32 bit. thus allowing for more address then IPv4. after that there nothing really interesting to talk about. 

Exactly what I wanted to say. Of course there are some other differences too, but those would be just way too technical for a Techquickie video. IP in general- maybe, IPv6-definitely not.

[LAwLz]

IPv6 is nearly exactly like IPv4. the only major difference is it uses 128 bit address instead of 32 bit. thus allowing for more address then IPv4. after that there nothing really interesting to talk about.

It's not that simple. More addresses is one of the biggest benefits but we also get stuff like:

 

1) Cleaner addressing and more efficient routing - What I mean by that is that we can go back to somewhat classed networks. Because there are so many addresses, each ISP can get their own prefix and only announce that address. This will not only help with performance (only send out 1 address instead of maybe 50) but also it will look a lot cleaner. So all AT&T customers might have a specific starting pattern for their IP, and all Comcast customers might have a specific starting pattern. This means that there is less computing needed to be done by routers in order to determine where a packet should be sent.

This, by the way, assumes that people are smart when designing their networks and implement their routing protocols though. Also, until we have completely replaced IPv4 we will have MORE addresses to keep track of, and therefore less efficient routing and it is less clean.

 

2) Simpler header - IPv6 has a larger header (because the addresses are bigger) but they actually require a lot less computation to be done because they are more simple.

IPv4 header has the following fields:

• Version
• Internet Header Length
• Type of Service
• Total Length
• Identification
• Flags
• Fragment Offset
• Time to Live
• Protocol
• Header Checksum
• Source Address
• Destination Address
• Options
• Padding

IPv6 has these fields:

• Version
• Traffic Class
• Flow Label
• Payload Length
• Next header
• Hop Limit
• Source Address
• Destination Address
• (Optional) Extension Header(s)

Green = Kept in IPv6

Red = Removed

Yellow = Changed but retains some or all of the functionality.

Purple = Now uses a different format, strict rules on how to use them and are also optional (don't need to be included in each packet) and supports all the important functions from IPv4, plus a few new ones (more on that later).

 

So as you can see, it is a lot more simple. The fields are also 64bit aligned which allows 64bit processors to process them faster (plus that they are more straight forward to process to begin with).

 

Found a good comparison picture:

 

3) No broadcasts - IPv6 does NOT support broadcasts. Instead we will use multicast far more often. Multicast already exists for IPv4 (although it was kind of glued on many years after IPv4 was invented) but IPv6 has made some changes to it and in general seems to have a more elegant implementation of it (not held together by glue). Sadly, I haven't looked into it so I can't really say much about it.

IPv6 also supports Anycast but I am not sure if IPv4 supports it as well. It seems like it is supported but barely anyone uses it, ever. IPv6 will probably use it a lot more.

 

 

4) Better auto-config - IPv6 supports something called Stateless Address Auto-configuration (SLAAC) which will remove the need for DHCP servers. How this work is that the router will advertise its prefix, and your device will auto generate the last 64 bits of the IPv6 address (the router also advertise the default gateway as well as how long the prefix is valid for) . One bad thing about this is that your device will use your MAC address for some of these 64 bits, which means that your public IP will also contain your MAC address. Embedding the MAC address is not mandatory (so you can inject whatever 64 bits you want but I suspect that the default state for most stuff will be to use the MAC address. This could be a pretty big privacy concern. You will even be able to tell which brand a computer has by only looking at the IPv6 address since this info is embedded.

If you don't want your MAC address to be embedded in your IPv6 address you have to enable the "Privacy Extension". What this will do is use a random number instead of your MAC address (it also has mechanics for avoiding two users getting the same address). But like I said before it will probably not be enabled by default, and the more things like that you enable, the more suspicious you look.

DHCP will still exist for IPv6 by the way. Not everyone will need it though.

 

 

5) Improvements to ICMP - This is far too much to write in a single post, and a lot of the details are so technical not even I understand them. Just comparing NDP against ARP is a lot. Wikipedia has a lot of details about ICMPv6 if you are interested in more details.

The summary is that a lot of basic functions needed have been changed/improved. SLAAC which I mentioned above is possible because of these changes.

 

 

6) IPsec built in, and better security in general - If you've ever looked into VPNs you have probably heard of IPsec. It's a framework used by things such as VPNs. It defines how encrypted packets should look. IPsec was originally developed for IPv6 but was then ported to IPv4 as well. All IPv6 devices will be able to securely send data between each other using IPsec. The bigger address space will also help defend against things such as people scanning for your IP simply because it is impractical to scan so many addresses.

This is one of the optional headers you can add.

7) No need for NAT and port forwarding - This comes back to the larger address space again. Right now we use NAT to map the same IP to multiple computers. This results in a lot of issues (like the need for port forwarding) which simply won't exist if we give each computer a unique address. It also takes some work off the router (no longer need to keep track of as many ports and addresses).

 

 

These are only some of the changes, and I only explained the basics of them. You can write a whole book about SLAAC alone, for example how it can use something called SeND.

 

 

TL:DR

I don't think a Fast as Possible (from now on called FAP) is a good idea on IPv6 because they are meant to be quick summaries of common things explained in an easy way. Doing an episode on IPv6 would be like firing a fire hose towards something that wants a glass of water.

Fapping IPv6 would either be far to quick and not thorough enough (summarize it as "it has more addresses and is better"), or it would be overwhelming for 99.9% of the viewers. It would probably just leave people unsatisfied.

Making one about IP and/or basic networking in general, and then quickly mention IPv6 and the expanded address space would work though. I don't watch the FAP videos so I don't know if he has already done this or not.