[Spce-user] peering weights possibly inverse in w.8 LTS

Matthew Ogden matthew at tenacit.net
Wed Nov 4 10:37:39 EST 2015


Hi All

according to documentation (and several threads I've looked at)

The selection of peering servers for outbound calls is done in the
following order: 1. length of the matching peering rules for a call. 2.
priority of the peering group. 3. weight of the peering servers in the
selected peering group. After one or more peering group(s) is matched for
an outbound call, all servers in this group are tried, according to their
weight (lower weight has more precedence). If a peering server replies with
SIP codes 408, 500 or 503, or if a peering server doesn’t respond at all,
the next peering server in the current peering group is used as a fallback,
one after the other until the call succeeds. If no more servers are left in
the current peering group, the next group which matches the peering rules
is going to be used.

"Lower has more precedence" for weight.

I changed my weights for some peering servers. 3 months ago.

They dont have the same prefixes, but some of their prefixes are the same
(mobile networks).

The have the same priority, but one peer has weight =1, the other peer has
weight =2.

After changing the 2nd peer to weight 2, I was expecting to see less mobile
calls made to it.

But since doing that, I'm seeing the opposite, (in both duration and number
of calls), peering with higher weight is getting more calls.

Is the documentation wrong, or is something else the issue?

Kind Regards

-- 





*Matthew Ogden*

Management

TenacIT





*Strategic IT Consulting *•* Advanced Networking *• *Virtualisation*

*Custom Development *• *Hosting *• *Syspro Support  *• *MS Licensing*

National Tel: 041 10 10 100 | Cell: 084 205 4445 | Email:
matthew at tenacit.net

CT Tel: 021 201 0333 | Skype Name: matthew.ogden | Web:
http://www.tenacit.net
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.sipwise.com/mailman/private/spce-user_lists.sipwise.com/attachments/20151104/6ae0fa13/attachment.html>


More information about the Spce-user mailing list