Wednesday, September 25, 2013

9971/9951/8945 and other video phones Exchange Unified Messaging fast busy/busy tone/not working

Hi Guys

As per the subject and mostly for me/you as well, when working with Exchange Unified Messaging and certain models of phones you might have issues with a fast busy tone when trying to check the voicemail.

Easy answer to this: you are probably enabled for Video, for some reason if you try and negotiate a video codec with Exchange 2010 it will fast busy on you, so either disable video on the phone OR create a region for the Exchange UM SIP trunk so that it doesn't use video.

Wednesday, September 4, 2013

ARC and migrated common phone profile

So if you upgrade ARC, be careful that your voice ports and CTI Route points don't have a "migrated common phone profile" automatically created and assigned as part of the upgrade process, because this can make them only talk IPv6 and break ARC (they won't register)

Shiney New Stuff in Vmware vSphere 5.5 and it's effect on your Cisco Enviroment

Hi Guys!

Lots of news about the new release of VMware vSphere 5.5, this blog post will focus purely on the vSwitch enhancements that I find most relevant to Cisco UCS and the Data Centre side, but hopefully a lot of it will apply for you in your Non-Cisco world too :) (Shout out to a certain Juniper lover I know ;))

Everything I am writing about here i took from the official VMware vSphere Announcement available here:

I won't be dealing with any of the SDN aspects of this announcement.

OK first thing's first, check out this feature:

Traffic Filtering
This feature allows you to specify ACL's based on MAC Address, System Traffic qualifications, and IP qualifiers (I.e Source/Dest Address, is the traffic vMotion Traffic?, what port number? etc.)

You can then filter the traffic (in any direction you need to), or more importantly, mark them with a DSCP or CoS Value, this is a crucial feature.

Take for example Cisco Communications Manager running on UCS: You create the vNIC, you assign CUCM the vNIC and it boots, when it sends out traffic, it will send them out with a DSCP value, which UCS cannot recognise: UCS can only recognise CoS Values.

So you can't use "Host Control" in your QoS policy on Cisco UCS, what you could potentially do is mark all traffic from CUCM with a particular CoS value, now the question is, What Cos Value? CoS 5? Now non-critical traffic is being put in a priority queue. Cos 3? Now your priority voice traffic for media services like conference bridges or annunciator is NOT priority.

Just one example of the problems you might have, another example might be with Exchange Unified Messaging: Same problem would exist, you would have to mark all traffic as a particular CoS value.

Previously the way to resolve this was to use Nexus 1000V, but now you have a way to do it in VMware which is great.

LACP Enhancements
There was a common misunderstanding with ESX  that not supporting LACP meant it only supported "cisco priopriatery etherchannels", which was just not true, PAGP and LACP are NOT etherchannel, they are just the advertisement protocol.  So you could always use a manual etherchannel (i.e. a static etherchannel that is not advertised) in the previous versions of ESX.

Anyway, they did introduce LACP support to make adding an etherchannel easier, but it had a restriction of only one LAG (one etherchannel essentially) per host per switch, this has now been changed so you can have up to 64 per host.

There is also some improvements to the load balancing algorithims used on ESX, apparently there are 22 new hashing algorithims available, which to me seem's very high, I will need to see what these are and the effect they may have on particular topologies, for now with Cisco UCS split between Fabric A and Fabric B keep using Load based on virtual port ID.

There are also some improvements to SRV-IO support but i don't have enough detail on this to be able to comment intelligently, I need to get a hold to see how this will help us with Adapter FEX and VM-Fex, my best guess at the moment is that the new enhancements for SRV-IO will allow you to use something very similiar to VM-FEX but natively within VMWARE.

Enhanced Host-Level Packet Capture
We are getting some major improvements in packet capturing on ESX, now we can capture straight on the command line of ESX (akin to tcpdump on a Linux system) You can now capture down to the vNIC level.

40GB NIC Support
There is now support for 40 gig nic's. This should hopefully stop 40 gig NIC's showing up as 4 seperate NIC's

I hope this has been interesting :).