Disadvantages/Advantages of IPv6 in Linux

Hello,
Can anybody give a short list of Disadvantages/Advantages of IPv6 in
Linux?
(comparing to IPV4).

I am more interested in disadvantages (because I think I know more
about the
advantages).
The only disadvantages I could think of are only these 3:

1) Currently there is no support to LVS in the linux kernel.

2) You need to port applications to IPV6 .

3) You need tunnels when you are working against IPV4 machines.


I would appreciate if anybody can add to this list of
disadvantages.

Regards,
Ian

<(E-Mail Removed)> wrote in message
news:642572fc-df8f-4441-aa26-(E-Mail Removed)...
> Hello,
> Can anybody give a short list of Disadvantages/Advantages of IPv6 in
> Linux?
> (comparing to IPV4).
>
> I am more interested in disadvantages (because I think I know more
> about the
> advantages).
> The only disadvantages I could think of are only these 3:
>
> 1) Currently there is no support to LVS in the linux kernel.

What is LVS?

> 2) You need to port applications to IPV6 .

Many applications are already converted, but are often pre-compiled without
such support. If you compile what you use from source, that's not a
problem.

> 3) You need tunnels when you are working against IPV4 machines.

That will eventually change. Tunnels aren't so difficult. I've used them
for 3+ years.

> I would appreciate if anybody can add to this list of
> disadvantages.

Advantage: No port scanning as the address space is really too big for such
to be practical. My firewall throws away about 30MB of packets per week of
port scans and other intrusion attempts to services I don't run.

However, also this week, there was a single attempt to spam over IPv6 (it
was "successful" for the spammer, but he spammed a spamtrap). Maybe it's
time to start an IPv6 based DNSBL?

(E-Mail Removed) wrote:
> Hello,
> Can anybody give a short list of Disadvantages/Advantages of IPv6 in
> Linux?
> (comparing to IPV4).
>
> I am more interested in disadvantages (because I think I know more
> about the
> advantages).
> The only disadvantages I could think of are only these 3:
>
> 1) Currently there is no support to LVS in the linux kernel.

>
> 2) You need to port applications to IPV6 .
>
> 3) You need tunnels when you are working against IPV4 machines.

Can't IPV4 be encapsulated in IPV6?
I thought you only needed tunnels to pass through routers, but not so
much to directly talk to IPV4 machines.
>
>
> I would appreciate if anybody can add to this list of
> disadvantages.

You should concentrate more on the advantages, like better performance,
less tweaks, no more need for NAT (jeej), more available IP's.

But one of the major disadvantages on IPV6 in general is that a lot of
sysadmins and people alike do not understand how it works... It
incorporates a learning curve.

>
> Regards,
> Ian
>

-R-

> like better performance

Can you elaborate a bit ? (except of eliminating the need for NAT).
There is something which I do not understand:
For routers in the internet - is the kernel routing table more
efficient than in IPV4?
As far as I understand, the routing is the same as in IPV4 (namely, we
find the route using
a prefix in IPV6 whrereas we have the subnet in IPv4).

I will appreciate if someone can elaborate on this.
Regards,
Ian

On Apr 2, 11:53 am, Jurgen Haan <jur...@fake.tld> wrote:
> ian...@gmail.com wrote:
> > Hello,
> > Can anybody give a short list of Disadvantages/Advantages of IPv6 in
> > Linux?
> > (comparing to IPV4).
>
> > I am more interested in disadvantages (because I think I know more
> > about the
> > advantages).
> > The only disadvantages I could think of are only these 3:
>
> > 1) Currently there is no support to LVS in the linux kernel.
>
> > 2) You need to port applications to IPV6 .
>
> > 3) You need tunnels when you are working against IPV4 machines.
>
> Can't IPV4 be encapsulated in IPV6?
> I thought you only needed tunnels to pass through routers, but not so
> much to directly talk to IPV4 machines.
>
>
>
> > I would appreciate if anybody can add to this list of
> > disadvantages.
>
> You should concentrate more on the advantages, like better performance,
> less tweaks, no more need for NAT (jeej), more available IP's.
>
> But one of the major disadvantages on IPV6 in general is that a lot of
> sysadmins and people alike do not understand how it works... It
> incorporates a learning curve.
>
>
>
> > Regards,
> > Ian
>
> -R-

(E-Mail Removed) wrote:
>> like better performance
>
> Can you elaborate a bit ? (except of eliminating the need for NAT).
> There is something which I do not understand:
> For routers in the internet - is the kernel routing table more
> efficient than in IPV4?
> As far as I understand, the routing is the same as in IPV4 (namely, we
> find the route using
> a prefix in IPV6 whrereas we have the subnet in IPv4).
>
> I will appreciate if someone can elaborate on this.
> Regards,
> Ian
>

IPV6 supports larger datagrams which will result in a (slightly) higher
troughput on highspeed networks (due to less overhead).

Jurgen Haan <(E-Mail Removed)> wrote:
> IPV6 supports larger datagrams which will result in a (slightly)
> higher troughput on highspeed networks (due to less overhead).

I suspect that requires a correspondingly larger MTU. I can get to
almost 64KB with IPv4 today (on agreeable hardware).

If I have an MTU <= 64KB, (eg 1500) then the larger header size of
IPv6 means a corresponding loss in maximum throughput since the ratio
of data to data+headers gets worse for IPv6 than IPv4. I've seen this
regularly with 1500 byte MTUs and netperf TCP_STREAM tests.

Also, (and some might consider this a good thing I guess) not as many
NICs offer stateless offloads like CKO, TSO and LRO for IPv6, so one
sees a correspondingly higher service demand (CPU consumed per unit of
data transferred).

Still, speaking for myself at least, I look forward to wider
deployment of IPv6 as I dislike NAT with a passion. That reminds me,
I need to pester my broadband provider about lighting-up IPv6... last
time I asked the reply made it sound like their support organization
thought they had to go to 64 bit address space to use IPv6...

rick jones
--
No need to believe in either side, or any side. There is no cause.
There's only yourself. The belief is in your own precision. - Jobert
these opinions are mine, all mine; HP might not want them anyway...
feel free to post, OR email to rick.jones2 in hp.com but NOT BOTH...

Hello,

Rick Jones a écrit :
> Jurgen Haan <(E-Mail Removed)> wrote:
>
>>IPV6 supports larger datagrams which will result in a (slightly)
>>higher troughput on highspeed networks (due to less overhead).
>
> I suspect that requires a correspondingly larger MTU. I can get to
> almost 64KB with IPv4 today (on agreeable hardware).

Besides, AFAIK IPv6 does not support datagrams larger than 65535 either.