Dear draft-ietf-opsawg-large-flow-load-balancing authors,
I started to AD review of draft-ietf-opsawg-large-flow-load-balancing
(Sorry for the delay btw).
Then I reviewed the WG LC feedback (email from Melinda, "[OPSAWG] WG
last call for "Mechanisms for Optimal LAG/ECMP Component Link
Utilization in Networks", on August 24th), and I realized that I have
the same feedback as Dan Romascanu in
http://www.ietf.org/mail-archive/web/opsawg/current/msg02891.html
Then I reviewed all the WGLC feedback, and it appears that this document
was sent on my plate, while the most up to date draft version 5 still
doesn't address the WGLC feedback.
For example:
http://www.ietf.org/mail-archive/web/opsawg/current/msg02897.html
"It refers to all of them, i.e. LAG, ECMP, and even the
(hierarchical) combination. We can try and provide some
clarification around that. We will add 802.1AX as a reference for
LAG."
Please post a revised version that addresses the WGLC comments.
While you are at it (and since I reviewed 1/3 of the document), here is
part of my feedback.
I'll restart my AD review with the next version
Regards, Benoit
=================================================================
- You use the term flow is a lot. Your definition is in line with the Flow
definition in RFC 7011 (IPFIX).
Why not reuse it?
Even your specific flow definition could re use the same "Flow" definition from
RFC 7011
Large Flow(s): long-lived large Flow(s)
Small Flow(s): long-lived small Flow(s) and short-lived small/large Flow(s)
- How many flow categories do you want to stress?
1. Introduction:
Network traffic can be predominantly categorized intotwo traffic
types: long-livedlarge flows andother flows (which includelong-
lived small flows, short-lived small/large flows)
1.2. Terminology
Large flow(s): long-lived large flow(s)
Small flow(s): long-lived small flow(s) and short-lived small/large
flow(s)
2. Flow Categorization
In general, based on the size and duration, a flow can be categorized
into any one of the followingfour types, as shown in Figure 1:
(a) Short-Lived Large Flow (SLLF),
(b) Short-Lived Small Flow (SLSF),
(c) Long-Lived Large Flow (LLLF), and
(d) Long-Lived Small Flow (LLSF).
Please be consistent. This is slightly confusing.
This only makes sense with the sentence at the end of section 2.
In this document, we categorize Long-lived large flow(s) as "Large"
flow(s), and all of the others -- Long-lived small flow(s) and short-
lived small/large flow(s) as "Small" flow(s).
Proposal (to make it clear earlier in document):
OLD:
Network traffic can be predominantly categorized into two traffic
types: long-lived large flows and other flows (which include long-
lived small flows, short-lived small/large flows)
NEW:
Network traffic can be predominantly categorized into two traffic
types: long-lived large flows and other flows.These other flow,
which include long-lived small flows and short-lived small/large flows,
are considered as small flows in this document.
EDITORIAL
-
Please expand LAG/ECMP in the abstract. While ECMP is known, LAG is not.
-
(load/mix and comma)
OLD:
If the traffic_load_constitutes flows such that the result of the
hash function across these flows is fairly uniform so that a similar
number of flows is mapped to each component link, if_,_ the individual
flow rates are much smaller as compared to the link capacity,
NEW:
If the traffic_mix_constitutes flows such that the result of the
hash function across these flows is fairly uniform so that a similar
number of flows is mapped to each component link, if the individual
flow rates are much smaller as compared to the link capacity,
-
EDITORIAL
OLD:
Where: ->-> small flows
===> large flow
NEW:
Where: -> small flow
===> large flow
Alternatively, make it clear in the figure that "-> -> ->" means 3 small flows,
by having each flow on a new line.
In the same figure, what does "--/---/-" mean?
-
To demonstrate the limitations of local optimization, consider a two-
level fat-tree topology with three leaf nodes (L1, L2, L3) and two
spine nodes (S1, S2) and assume all of the links are 10 Gbps. Let L1
have two flows of 4 Gbps each towards L3, and let L2 have one flow of
7 Gbps also towards L3. If L1 balances the load optimally between S1
and S2, and L2 sends the flow via S1, then the downlink from S1 to L3
would get congested resulting in packet discards. On the other hand,
if L1 had sent both its flows towards S1 and L2 had sent its flow
towards S2, there would have been no congestion at either S1 or S2.
A figure would be appreciated.
- as a courtesy for the RFC editors, please order the references
- idnits complains about the date.
- RFC 5101 -> RFC 7011
_______________________________________________
OPSAWG mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/opsawg
