Re: [ipv6-wg] ipv6-wg Digest, Vol 55, Issue 2

[Benedikt Stockebrand]

Hi Jens and list,

Jens Link  writes:

> Benedikt Stockebrand  writes:
>
>> They used AWS/S3 for some relevant stuff, and since it was done
>> externally it wasn't properly QAed.  When Amazon switched IPv6 off
>> again, they had a little bit of an issue.  We only found out kind of
>> accidentially, especially so because they didn't want to make it all
>> that obvious that they are using Amazon.
>
> Can't be that relevant if it was not monitored properly.

sorry, but I really can't publicly get into the details of that.  Let's
just say this was the tip of the iceberg, or the trailer of the TV
series, or the reason I got so fond of drain cleaner...


Cheers,

Benedikt

-- 
Benedikt Stockebrand,   Stepladder IT Training+Consulting
Dipl.-Inform.   http://www.stepladder-it.com/

  Business Grade IPv6 --- Consulting, Training, Projects

BIVBlog---Benedikt's IT Video Blog: http://www.stepladder-it.com/bivblog/

Re: [ipv6-wg] ipv6-wg Digest, Vol 55, Issue 2

[Jen Linkova]

On Thu, May 5, 2016 at 1:11 PM, Jens Link  wrote:
> Benedikt Stockebrand  writes:
>
>> They used AWS/S3 for some relevant stuff, and since it was done
>> externally it wasn't properly QAed.  When Amazon switched IPv6 off
>> again, they had a little bit of an issue.  We only found out kind of
>> accidentially, especially so because they didn't want to make it all
>> that obvious that they are using Amazon.
>
> Can't be that relevant if it was not monitored properly.

Your statement is true - but in the ideal world only...


-- 
SY, Jen Linkova aka Furry

Re: [ipv6-wg] ipv6-wg Digest, Vol 55, Issue 2

[Jens Link]

Benedikt Stockebrand  writes:

> They used AWS/S3 for some relevant stuff, and since it was done
> externally it wasn't properly QAed.  When Amazon switched IPv6 off
> again, they had a little bit of an issue.  We only found out kind of
> accidentially, especially so because they didn't want to make it all
> that obvious that they are using Amazon.

Can't be that relevant if it was not monitored properly.

Jens
-- 

| Foelderichstr. 40   | 13595 Berlin, Germany   | +49-151-18721264 |
| http://blog.quux.de | jabber: jensl...@quux.de| ---  | 

Re: [ipv6-wg] ipv6-wg Digest, Vol 55, Issue 2

[Jens Link]

Benedikt Stockebrand  writes:

Hi,

>> Well many content providers / startups use "the cloud". No IPv6 there,
>> no content. 
>
> Stopitstopitstopitstopit!

Seriously. Several people told me "If AWS doesn't offer IPv6 we don't
need IPv6." On the other hand someone told me "our solution for IPv6
hosting is Amazon. Or some other cloud provider."

Jens 
-- 

| Foelderichstr. 40   | 13595 Berlin, Germany   | +49-151-18721264 |
| http://blog.quux.de | jabber: jensl...@quux.de| ---  | 

Re: [ipv6-wg] ipv6-wg Digest, Vol 55, Issue 2

[Benedikt Stockebrand]

Hi Jens and list,

Jens Link  writes:

> Benedikt Stockebrand  writes:
>
> Hi,
>
>>> And may I add cloud providers?
>>
>> No, you may not.  Definitely not.  Go away.  And take those enterprises
>> using them as a cheap CDN with you...
>
> Well many content providers / startups use "the cloud". No IPv6 there,
> no content. 

Stopitstopitstopitstopit!

>> And what's even more frustrating: The Amazon stuff at some time
>> supported IPv6 at least on a best effort base, but they switched it off
>> again.
>
> AFAIK only for HTTP(S) Loadbalancing.

Which according to some totally unconfirmed rumors was "good enough" for
some organization to use as their wannabe CDN.  And then Amazon switched
IPv6 support off again...


I want a drink.  And something strong.  Like drain cleaner, or at least
battery acid.


Cheers,

Benedikt

-- 
Benedikt Stockebrand,   Stepladder IT Training+Consulting
Dipl.-Inform.   http://www.stepladder-it.com/

  Business Grade IPv6 --- Consulting, Training, Projects

BIVBlog---Benedikt's IT Video Blog: http://www.stepladder-it.com/bivblog/

Re: [ipv6-wg] ipv6-wg Digest, Vol 55, Issue 2

[Jens Link]

Benedikt Stockebrand  writes:

Hi,

>> And may I add cloud providers?
>
> No, you may not.  Definitely not.  Go away.  And take those enterprises
> using them as a cheap CDN with you...

Well many content providers / startups use "the cloud". No IPv6 there,
no content. 

> And what's even more frustrating: The Amazon stuff at some time
> supported IPv6 at least on a best effort base, but they switched it off
> again.

AFAIK only for HTTP(S) Loadbalancing.

Jens
-- 

| Foelderichstr. 40   | 13595 Berlin, Germany   | +49-151-18721264 |
| http://blog.quux.de | jabber: jensl...@quux.de| ---  | 

Re: [ipv6-wg] ipv6-wg Digest, Vol 55, Issue 2

[Jen Linkova]

On Tue, May 3, 2016 at 12:37 AM, Benedikt Stockebrand
 wrote:
>>> That would be a great panel discussion with some diverse speakers on the 
>>> panel  :-)
>>
>> I have been doing some enterprise stuff as well recently. If there is
>> going to be such a panel I would love to participate! :)
>
> Unless Jen somehow scares away various speakers (and I guess that's
> something she's *not* going to be particularly successful with:-) our
> schedule is already pretty full.
>
> Otherwise, I heartily agree.  IPv6 is currently changing from an ISP
> issue to an enterprise and (especially small to medium) content provider
> issue.  And if I've learned anything the last two years, then that this
> opens a completely different can of worms.
>
> Maybe we can do that panel discussion in Madrid(?)

Sounds like a plan ;)

-- 
SY, Jen Linkova aka Furry

Re: [ipv6-wg] ipv6-wg Digest, Vol 55, Issue 2

[Sander Steffann]

Hi,

> Op 25 apr. 2016, om 19:35 heeft Silvia Hagen  het 
> volgende geschreven:
> 
> That would be a great panel discussion with some diverse speakers on the 
> panel  :-)

I have been doing some enterprise stuff as well recently. If there is going to 
be such a panel I would love to participate! :)

Cheers,
Sander



signature.asc
Description: Message signed with OpenPGP using GPGMail

Re: [ipv6-wg] ipv6-wg Digest, Vol 55, Issue 2

[Silvia Hagen]

That would be a great panel discussion with some diverse speakers on the panel  
:-)

Silvia

-Ursprüngliche Nachricht-
Von: ipv6-wg [mailto:ipv6-wg-boun...@ripe.net] Im Auftrag von Benedikt 
Stockebrand
Gesendet: Montag, 25. April 2016 20:14
An: christian bretterhofer
Cc: ipv6-wg@ripe.net
Betreff: Re: [ipv6-wg] ipv6-wg Digest, Vol 55, Issue 2

Hi Christian and list,

christian bretterhofer <christian.bretterho...@gmail.com> writes:

> I think the basic work for ISPs in concern to IPv6 is covered.

well, depends on the ISP in question.  To me it looks a lot like many are still 
struggling to get the necessary knowledge and experience to their tech and 
support crowd---not necessarily with the people actively involved in the RIPE 
community, but at least with the big ones.

A customer recently asked one of the large players here in Germany if they were 
interested in a contract that would have allowed my customer to outsource some 
IPv6-related tasks---or rather, to outsource some tasks that were also expected 
to be supported via IPv6.  They were turned down with the explanation "we don't 
have the necessary manpower to operate this".

> But i miss the topics to be addressed if you want to migrate from a
> IPv4 Microsoft Active domain using company to an system where most 
> server in an enterprise could by just IPv6 only and use technologies 
> like NAT46 ( SIIT-DC ) or similar to still make IPv4 only windows 
> clients happy.

Now I've taken a bit of a look at these things, but then I'm not exactly a 
Microsoft guy.  From all I've seen, going for NAT64 and such is generally a bad 
idea.  Instead, ensure that IPv6 is provided wherever it is needed and then 
make your servers dual stacked.

Yes, that frequently involves upgrades on various servers nobody really wants 
to touch, but the very reasons why nobody wants to touch them are the reasons 
why you actually clean that stuff up.

> Switching an enterprise with location around the global from a "we 
> donot route any IPv6 traffic across our WAN Links" "most servers have
> IPv6 disabled" to
> We start IPv6 routing partially and enable partial IPv6 support on 
> servers in a Microsoft ADS environment seems not covered in most IPv6 
> covering websites and presentations.

That may be because your approach is unnecessarily painful.  You want to get 
IPv6 up and running in the network infrastructure first, then make your servers 
dual-stacked and then deal with the clients.

At least that's the "strategic" outline of an approach.  Beyond that it's 
really a lot of detail work to do on an individual basis.

> Maintaining dual stack for the datacenters is just painfull and there 
> should be a "single" device in the form of NAT46/SIIT/SIIT-DC in front 
> of each server area. I am not sure that Active directory is ready for 
> that.

Nonononono, don't do that.  Whenever something goes wrong with that "single 
device", you'll have a serious disruption of service, not everything works 
through it, and you'll never ever get a chance to get rid of it in the long run 
because there'll always be that one last server that depends on it, or might 
depend on it but nobody knows for sure.

Yes, that means that you need to have all your servers dual stacked, and yes, 
that's some serious extra workload in a data center context, but anything else 
is quite likely way worse.


Cheers,

Benedikt

-- 
Benedikt Stockebrand,   Stepladder IT Training+Consulting
Dipl.-Inform.   http://www.stepladder-it.com/

  Business Grade IPv6 --- Consulting, Training, Projects

BIVBlog---Benedikt's IT Video Blog: http://www.stepladder-it.com/bivblog/

Re: [ipv6-wg] ipv6-wg Digest, Vol 55, Issue 2

[Benedikt Stockebrand]

Hi Christian and list,

christian bretterhofer  writes:

> I think the basic work for ISPs in concern to IPv6 is covered.

well, depends on the ISP in question.  To me it looks a lot like many
are still struggling to get the necessary knowledge and experience to
their tech and support crowd---not necessarily with the people actively
involved in the RIPE community, but at least with the big ones.

A customer recently asked one of the large players here in Germany if
they were interested in a contract that would have allowed my customer
to outsource some IPv6-related tasks---or rather, to outsource some
tasks that were also expected to be supported via IPv6.  They were
turned down with the explanation "we don't have the necessary manpower
to operate this".

> But i miss the topics to be addressed if you want to migrate from a
> IPv4 Microsoft Active domain using company to an system where most
> server in an enterprise could by just IPv6 only and use technologies
> like NAT46 ( SIIT-DC ) or similar to still make IPv4 only windows
> clients happy.

Now I've taken a bit of a look at these things, but then I'm not exactly
a Microsoft guy.  From all I've seen, going for NAT64 and such is
generally a bad idea.  Instead, ensure that IPv6 is provided wherever it
is needed and then make your servers dual stacked.

Yes, that frequently involves upgrades on various servers nobody really
wants to touch, but the very reasons why nobody wants to touch them are
the reasons why you actually clean that stuff up.

> Switching an enterprise with location around the global from a "we
> donot route any IPv6 traffic across our WAN Links" "most servers have
> IPv6 disabled" to
> We start IPv6 routing partially and enable partial IPv6 support on
> servers in a Microsoft ADS environment seems not covered in most IPv6
> covering websites and presentations.

That may be because your approach is unnecessarily painful.  You want to
get IPv6 up and running in the network infrastructure first, then make
your servers dual-stacked and then deal with the clients.

At least that's the "strategic" outline of an approach.  Beyond that
it's really a lot of detail work to do on an individual basis.

> Maintaining dual stack for the datacenters is just painfull and there
> should be a "single" device in the form of NAT46/SIIT/SIIT-DC in front
> of each server area. I am not sure that Active directory is ready for
> that.

Nonononono, don't do that.  Whenever something goes wrong with that
"single device", you'll have a serious disruption of service, not
everything works through it, and you'll never ever get a chance to get
rid of it in the long run because there'll always be that one last
server that depends on it, or might depend on it but nobody knows for
sure.

Yes, that means that you need to have all your servers dual stacked, and
yes, that's some serious extra workload in a data center context, but
anything else is quite likely way worse.


Cheers,

Benedikt

-- 
Benedikt Stockebrand,   Stepladder IT Training+Consulting
Dipl.-Inform.   http://www.stepladder-it.com/

  Business Grade IPv6 --- Consulting, Training, Projects

BIVBlog---Benedikt's IT Video Blog: http://www.stepladder-it.com/bivblog/