[c-nsp] Products for dealing with packet re-ordering?

[James Bensley]

On 9 January 2018 at 15:51, Alex K. <nsp.lists at gmail.com> wrote:
> Hello everyone,
>
> An interesting issue came across my desk, one I'd like to hear community
> opinion about.
>
> A customer of mine, due to carrier restrictions, is about to aggregate a
> few links into one (using etherchannel). Due to required bandwidth, it
> seems he would need to load balance across the LAG on a packet by packet
> basis. Now, we all know the dangers of packet reordering. On another hand,
> there are products from VeloCloud, Talari, SilverPeak and probably many
> others, that on paper are capable of dealing with the issues and delivering
> packets in orderly manner.
>
> The question is, is anybody using those products and can admit first-hand
> experience? Or the daemon of packet reordering is grossly exaggerated?
>
> Thank you for sharing your experience/opinions on the matter.

What bandwidths and connection types are you referring too?

You mentioned VeloCloud and SilverPeak, does this mean you looking at
some SD-WAN solutions specifically? Is this end site connectivity
specifically or not?

Both of the vendors above do handle packet miss ordering, we've been
working with both and they use sequence numbers on their respective
custom tunnelling protocol to re-order packets. They aim to service
scenarios where you have traffic being split over quite different
links e.g. an ADSL and Cable/Coax service in an active/active
per-packet load-balancing deployment.

We also use MLPPP to provide bonded ADSL which handles packet
reordering without any specialist software/hardware so if you are
looking at access layer stuff maybe that is an option to you. We also
provide EFM and EFM has packet re-ordering built in to the protocol as
it is effectively per-packet load balancing over SHDSL links.

However, if this is in the DC, if you customer is creating a LAG
between their DC switch and a carrier DC switch and you're proposing N
links between the same two devices in the same two racks, and you have
the exact same copper/fibre runs that are the same length etc., then I
would expect the miss-ordering to be minimal (not zero but likely an
amount deemed as acceptable). In this scenario with proper monitoring
to keep tabs on it, I would expect no special hardware required.

Cheers,
James.