<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=iso-8859-1">
<meta name="Generator" content="Microsoft Exchange Server">
<!-- converted from text --><style><!-- .EmailQuote { margin-left: 1pt; padding-left: 4pt; border-left: #800000 2px solid; } --></style>
</head>
<body>
<div>
<div dir="auto">If jitter were defined anywhere vis-à-vis LACP, it would be _de jure_, not _de facto_ as I said.
<div dir="auto"><br>
</div>
<div dir="auto">Yes, if you have *guaranteed* that TCP sessions hash uniquely to a single link in your network, you might be able to successfully tunnel LACP (or EtherChannel, or any other L1 link-bonding technique). The last time I attempted to do this on
my network, I discovered that guarantee wasn't nearly as ironclad as I expected. I don't remember the gory details, at this remove, sorry. Maybe it wasn't TCP? Maybe it wasn't the default hashing algorithm? Dunno.</div>
<div dir="auto"><br>
</div>
<div dir="auto">-Adam</div>
</div>
<div class="x_gmail_extra"><br>
<div class="x_gmail_quote">On Jul. 8, 2020 00:48, Saku Ytti <saku@ytti.fi> wrote:<br type="attribution">
</div>
</div>
</div>
<font size="2"><span style="font-size:10pt;">
<div class="PlainText">Hey Adam,<br>
<br>
On Wed, 8 Jul 2020 at 00:11, Adam Thompson <athompson@merlin.mb.ca> wrote:<br>
<br>
> Good luck with tunnelling LACP, no matter what boxes you have - LACP has (de facto) hard jitter requirements of under 1msec, or you'll be getting TCP resets coming out your ears due to mis-ordered packets.<br>
<br>
Can you elaborate on this? Where is LACP jitter defined and for what<br>
purpose? We push packets around the globe in sub 200us jitter on any<br>
given day, so 1000us isn't for us a particularly hard goal.<br>
<br>
Only reason why I could imagine someone would care about jitter here<br>
is if protocol measures delay (LACP doesn't) and relies on delay to<br>
remain static and then balances per-packet or per-byte or otherwise<br>
between multiple links.<br>
However we of course put all packets from given TCP session to always<br>
same LACP interface, so from TCP session POV, each LACP is exactly a 1<br>
interface. Per-packet balancing on LACP is possible via a special<br>
configuration, but anyone who does it, doesn't care about reordering,<br>
no matter of jitter, because even in very stable jitter, the paths may<br>
be unequal length and cause reordering.<br>
<br>
LACP hellos are sent every 1s when in fast mode with 3s keepalive,<br>
which also isn't particularly tight. We do have customers running LACP<br>
over MPLS pseudowires over great distances.<br>
<br>
-- <br>
++ytti<br>
</div>
</span></font>
</body>
</html>