[Measurement-wg] WhatsaApp traffic pattern for over the Internet

[Josiah Chavula]

Thank you,  Dr Nyirenda, you raise pertinent questions. I have an ongoing
interest in these, and would be happy to collaborate.

On Fri, Jul 24, 2020 at 10:57 AM Dr P Nyirenda <paulos at sdnp.org.mw> wrote:


> AF-IX,

>

> 3. Since the WhatsApp traffic can hence be considered local, does an IXP

> make any

> difference, can it be made to make a difference?

>


With WhatsApp's XMPP messages having to bounce-off WhatsApp servers, and
without a local FFBCDN node, I am not sure how the local IXP would help in
localizing WhatsApp traffic.



> 4. Is this WhatsApp traffic measurable so that we can learn its

> characteristics?

>


This is quite possible, I have seen a few academic papers on this. We (at
UCT) are doing something similar for wireless community network traffic.
With WhatsApp traffic being e2e encrypted, one needs a more careful
methodology to isolate Whatsapp flows, and one approach is to employ
deep-packet-inspection to determine WhatsApp/Facebook flows for analysis.
Happy to collaborate with you on this if we can obtain some data. The first
and obvious challenge I see is, would Malawi's MNOs/ISPs be willing to
grant us some anonymized flow-data?


>

> In Malawi there is currently an outcry over personal expenditures or costs

> over data bundles

> and hence there is a lot of interest in bringing such costs down. This is

> hence a pertinent

> issue that seems to need resolution.

>


In terms of further brining the costs down, I am hoping that our
operators/mispa/mixp still endeavour to attract major CDNs into the
country, particularly Google Global Cache (GGC) and FBCDN.


>

-- 
Josiah Chavula, PhD
Computer Science Department
University of Cape Town
Rondenbosch, Cape Town, South Africa
Tel: +27 21 650 1330
www.josiahchavula.com/
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.afrinic.net/pipermail/measurement-wg/attachments/20200726/0073a81a/attachment.html>