Saturday 27 July 2013

WebRTC - LTE - Over The Top (OTT) services and disruption - or opportunity?

I've recently been looking hard at IMS and LTE - as both these technologies are finally gaining traction in the market place with Carriers upgrading their networks to 4G with VoLTE to follow very soon - within a year or so for many.

I've also been closely following the rapid rise of WebRTC and its deployment in browsers, Firefox and Google Chrome on you desktop now support WebRTC APIs and very soon it will be on your smartphone (http://googlechromereleases.blogspot.co.uk/2013/07/chrome-beta-for-android.html).

Carrier vendors such as Genband and others are all ready with their implementations of App servers capable of providing WebRTC.

There is plenty of support in the Open Source community with Asterisk, Freeswitch and Kamailio and OpenSIPs support SIP over websockets and OverSIP (http://oversip.net/) and the sipML5 HTML5 (Javascript SIP/SDP stack)... Not to mention MobiCents Open source IMS app server's support for it (http://www.slideshare.net/johntel/crocodile-launch-mobicents-html5-web-rtc-and-sip-over-websockets).

More Interesting on this front as well is the recently announced Project Clearwater (http://www.projectclearwater.org/) IMS in the cloud.

It seems to me there is about a 1 year opening for the Over The Top providers (OTT) to get in on LTE and capitalise on the carrier's reluctance to use VoLTE. Its not to say there won't be issues - carrier's LTE implementations will probably be restricted to a single Data service offering for this first year, offering Dongles and WiFi Personal hotspots (MyFI) over a none QoS bearer service, which might not be the best for offering voice and video services (but that doesn't stop most of us using Skype!)

From the carriers perspective it might also be worth considering their position with respect to partnerships to enable OTT providers or deep packet inspection technologies to block the OTT vendors from utilising WebRTC applications. Partnership in my mind is better for both. It gives OTT providers the opportunity to get a QoS bearer service for their applications (QCI 5 and QCI 1 EPS Bearers) and potentially a custom APN to support this. For the Carriers innovation has not historically been a carrier's strength - partnership brings the opportunity to gain revenue from innovative services with little or no cost of development, and prevents them from the constant fear of becoming a bit-pipe provider only.

Whilst thinking about QoS/QoE it also struck me that its going to be a bit more tricky (not impossible) for 3rd party QoS/QoE software probes to monitor the Quality of Web RTC traffic, on two counts, 1) its peer to peer - unless forced to be otherwise (media anchoring); 2) the media and RTCP is "munged" (technical term) together into a single UDP stream to improve on the chances of successful NAT traversal. And on a third count - the media may also be encrypted.

Wednesday 17 July 2013

VMWare of Microsoft HyperV - Free or Licenced?

Whilst considering the design requirements of a hosted contact centre solution, one of the requests has been to consider which elements can be considered for virtualisation. The products in question for consideration are Oracle CallCallCenterAnywhere (CCA) and Cosmocom Call Universe (CCU).

Oracle CCA is officially end of life and now in extended support from Oracle - this doesn't stop people using it though and both my self and a number of consultants are still supporting this platform, testament to the guys at Telephony@Work who created this great product and shame on Oracle for end-of-life for the product. Fix Pack 10 supports Windows server 2008 - 32 bit only officially however the small print in the release notes indicates the executables have been tested to run on 64-bit Linux. This means there at least is the potential to support both HyperV and VMWare hypervisors.

That said the Cosmocom Call Universe product is now the only real alternative for multi-tenant host (Carrier scale) Contact Centre platform and thanks to the continued support and development fully support virtualisation. (footnote: I'm hoping for big things from Genesys Labs with their new cloud offering in this space).

Both CCA's and CCU's architecture support redundancy in their design, which means technologies such as live migration/vmotion are redundant in some respects for per data centre availability. However with 2 data centres the option for "WAN" live migration opens up the potential for a higher availability of the elements. This point requires careful consideration of the options. Since the other options open to is is just databased replication across the WAN to keep items in sync.

Having spent some time assessing VMWare vs. HyperV and product support of these hypervisors from Cosmocom - VMWare wins. Primarily on known to work on ESXi hypervisor, since CCU is currently only tested on Windows server 2008 with HyperV role. And secondary consideration is the footprint of the hypervisor with the the VMWare being a real light weight at around 512 Meg RAM!