Re: [Rift] Fw:New Version Notification for draft-zwx-rift-leaf-ring-00.txt

Hi Tony, 
Thank you very much for your comments!
Yes, we thought about the hierarchical solution before, but it's hard. 
In this figure I simplify the network, in the actual deployment, many nodes like L05 need to access the network. 
If we use the hierarchical levels, there may be many levels here and may exceed the max level limit. 
Do you have any other solution suggestion? Much appreciate for it. :-)
Best regards,
Sandy

------------------原始邮件------------------
发件人：TonyPrzygienda
收件人：张征00007940;
抄送人：rift@ietf.org;
日 期 ：2021年07月27日 03:37
主 题 ：Re: [Rift] Fw:New Version Notification for draft-zwx-rift-leaf-ring-00.txt
_______________________________________________
RIFT mailing list
RIFT@ietf.org
https://www.ietf.org/mailman/listinfo/rift

Read that with interest. RIFT is obviously quite applicable for the ring leaf stuff ;-) and I saw the use case before.

Simplest solution IMO (assuming you're running RIFT on the ring as your draft does) is to let the hierarchy figure out the CLOS variant here. ToFs start @ 24 per spec so there is plenty space if that's not enough it's easy to implement RIFT in a way where ToF start @ a higher number.

So to use your example you will end up with a topology that is a very fractioned CLOS ;-) but will work just fine given that L05 will get default from both L04 and L06. Observe that if you end up with L05 and L05' @ the same level you can enable horizontal link (leaf-2-leaf) but that won't do much
since default will point northbound and if L05 or L05' loses their northbound link they will just fall down in hierarchy to X-4 and use the other L05* as default

So I'm not sure what the draft would buy us over naive use of RIFT to stack more levels on top of each other. May be I'm missing something but we'll hear the preso I guess.

-- tony

|               |  |  |  |  |          | |   | | +----------------+  |  +-----------------+ | | |             |     |     |          | | | | | | +------------------------------------+ | | | | |           |     |     |          |   | |++-+-+--+        | +---+---+ |        +-+---+-++|       |        +-+       +-+        |        ||  L01  |--------- |  L02  |----------|  L03   |  Level X-1+---+---+          +-------+          +---+----+    |                                     |          |                                     |      ++-+-+--+                             ++--+---+ |       |                             |       |  Level X-2|  L04  |                             |  L06  |+---+---+                             +-+-----+    |                                   |    +----------------+  +---------------+                                  |  |                   ++-+-+--+                   |       |                   |  L05  |  Level X-3                   +-------+

On Thu, Jul 8, 2021 at 3:07 AM <zhang.zheng@zte.com.cn> wrote:
Hi,
A new draft for RIFT deployment and proposed extension is submitted.
Much appreciate if you spend some time to read this draft.
Any comments are welcome.
Thanks,
Sandy
------------------原始邮件------------------
发件人：internet-drafts@ietf.org
收件人：徐本崇10065053;徐本崇10065053;魏月华00019655;张征00007940;
日 期 ：2021年07月08日 08:48
主 题 ：New Version Notification for draft-zwx-rift-leaf-ring-00.txt
A new version of I-D, draft-zwx-rift-leaf-ring-00.txt
has been successfully submitted by Zheng Zhang and posted to the
IETF repository.
Name:        draft-zwx-rift-leaf-ring
Revision:    00
Title:        Supporting leaves without northbound neighbors connecting to a fat-tree network using RIFT
Document date:    2021-07-07
Group:        Individual Submission
Pages:        8
URL:            https://www.ietf.org/archive/id/draft-zwx-rift-leaf-ring-00.txt
Status:         https://datatracker.ietf.org/doc/draft-zwx-rift-leaf-ring/
Html:           https://www.ietf.org/archive/id/draft-zwx-rift-leaf-ring-00.html
Htmlized:       https://datatracker.ietf.org/doc/html/draft-zwx-rift-leaf-ring
Abstract:
This document discusses the usage and solution for leaf nodes without
northbound neighbors connecting to a fat-tree network by leaf nodes
having direct northbound neighbors in RIFT.
The IETF Secretariat_______________________________________________
RIFT mailing list
RIFT@ietf.org
https://www.ietf.org/mailman/listinfo/rift