[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [SIG-IX] IPv6 Peering
Hi,
On Mon, 17 Mar 2008 13:44:50 +0800
Raphael Ho <raphael.ho@ap.equinix.com> wrote:
> 1. Would you prefer:
> a. to use dual-stack on the same port/collision domain
> b. to have an entirely separate collision domain for v6 peering
a.
JPIX experimentaly started "stand alone v6 only separate port/collision
domain IX" from Jan. 2002. We chose this option because our policy was
"make it simple and stable, didn't wanted to affect IPv4 production
traffic." We were very conservative :-)
As I have announced on APNIC25 meeting, we decided starting
"dual-stack on the same port/collision domain" from Aug. 2008.
Now, we have tools to monitor v6 traffic using sflow...
# still Ishii-san, Mark-san, and Arien-san's point is true and
considerable.
Positive demand from members (like Ethern-san), started coming up.
# As Kato-san said, we did not have much demand of dual stack.
Trend was opposite from Arien-san's region.
To start dual-stack, traffic type could be separated using
Ethertype ACLs.
There are 3 types of traffic we need to think about,
1. v4 only
2. v6 only
Separating them using port/vlan base ether_type filter.
3. v4/v6 dual-stack
allow [1.2.] above or all.
# Still in need of testing, for reliablity and wirespeed
> 2. Would you prefer:
> a. bi-lateral peering
> b. multi-lateral peering
a.
But we will also offer a route arbitrator for a member who wants to
use it.
Best regards,
Takejiro Takabayashi(JPIX)
-------------------
Takejiro Takabayashi [Engineer]
Japan Internet Exchange Co, Ltd
Phone: +81-3-3243-9579
Fax: +81-3-3243-9597
E-mail: takejiro@jpix.ad.jp
