Metaswitch Forum 2014 will kick off soon, stay tuned.
Any second now - its 8:05 am in New Orleans
The event has kicked off with a keynote presented by Steve Gleave - which has been preceded by a few well-produced videos. The idea so far is tying in the theme of the event "The brains of the new global network" with Steve's brain. The cast of characters behind the scenes show how they tinker with Steve's brain - he eventually comes out on stage on a Segway with Harley Davidson sound-effects. And we are off.
Now to the theme of the presentation:
- The brains of the network are efficient and on the move - as smooth as data on a private cloud
- The network will have a new nervous system
- There is nothing like any of us - we are all unique, so are our customers and our approach to serving them
Steve went on to correlate the brain with network and to explain how evolution in the network - like the move to network functions virtualization or NFV. Media gateways, SDN and other technologies were woven into the brain discussion. From there there was talk of sensory maps in the brain which are roughly analogous to the functions in the carrier network.
Metaswitch CEO John Lazar takes the stage
John talked about the move to cloud, IP, software and a more open platform which will allow for things like fast-failure. IP interconnect is an example he cited. He went on to say networks need to be more progammable. He also said we need networks which ar ebetter at generating revenue. He continued by discussing OTT solutions like Skype and WhatsApp which are killing revenue. From there he discussed services like Netflix which can take up to 50% of the bandwidth with no corrsponding revenue.
He summarized by saying networks need to be more compeitive as they dont compete with the new entrants effectively.
Another point he made was the brains of the network on the move refers to the idea that network functions are moving to open systems such as OTS computers made by companies like Dell, etc.
He went on to say the data plane and media processinga are also moving to open systems. This has to work with a cloud orchestration layer in order to work in an elastic way.
Jason Silva , Futurist and Host of National Geographic's Brain Games takes the stage
Silva starts off by talking about how incredible technology is - how it evolves in an exponential fashion versus the linear evolution which our brains have experienced. Quoting Ray Jurzweil, he discusses the 30 linear step example which compares 30 linear and 3 0 exponential steps. The point is if you double the number each step you get to a billion in the exponential example.
Maybe the funniest line of the day so far:
If you dont have ADD today, you arent paying attention.
From here we started to watch some short video clips he produced.
Here is one example video called "Awe"
EarthLink is now 20 years old and the consumer business is less than 20%. Discussion of internal info - CEO change, the brand is trusted, etc. They will be investing to take advantage of this fact soon he says. Interesting - acquisitions perhaps?
Lots of discussion on product mix - wholesale, retail, MSP, cloud, etc.
Hosted voice is one of their fastest-grwoing products.
Went into a case study: Ciao Bella Gelato, had problems with email, small IT staff and too much time troubleshooting IT inffrastructure. EarthLink came in and put the voice and data in cloud - avoiding CAPEX and performance increased because equipment is connected to their network backbone. Three-weeks later super-storm Sandy took their HQ out of power for two weeks but their operations weren't affected. They would have been out of business for two-weeks he said.
Martin Taylor , CTO, Metaswitch takes the stage
Martin is setting up a talk which espouses the virtues of NFV - for more, check out my blog for stories on the topic and come to Software Telco Congress in Las Vegas this August. I expected a more technical deep-dive - interesting that we are getting an overview. I am sure the company knows what it is doing here as NFV is in th early-stages still and most carriers here are smaller meaning they aren't paying as much attention to the latest specifications and standards coming out of ETSI.
Having said that, the value prop being discussed is proprietary boxes are not only less powerful than OTS servers, they become boat anchors after a while. He says we are shacking network functions to a hardware replacement cycle. With the open-server model, the software lives on beyond the hardware cycle. Finally, he said, "You have disconnected your network function lifecycle from hardware relacement lifecycle."
He continued, saying software vendors have lower time-to-market than hardware companies and costs are lower as well. He then tied in the idea that Metaswitch is perfectly adapted to this world where telco expectations on software quality ae tremendous as we need to retain five-nines, etc.
Another point he made is software procurement and deployment should be much faster than hardware meaning less time to market for new and enhanced services.
Multi-tenancy is another area he focused on - explaining that carriers can sell wholesale services on virtualized systems which are isolated from one another.
Moreover, he says with NFV or a software-model, it is easier to customize services for particular customers far more easily.
He also said with virtualized solutions you can allow applications ot be installed in the IMS solution - or you can allow the enterprise to upload changes etc. The idea here isnt you arent risking your entire IMS operation if something goes wrong.
He also touched on power consumption - the idea being servers which aren't being used can be powered down... Think nights for example. He then went on to discuss how orchestration needs to handle this sort of thing. Here are the specific things he mentioned at this point:
- One click service deployment
- Composition of complex service chains
- Automated montitoring and repair
- Automatic dynamic scaling
- Automatic optimization of resources
Service chains can be customized to meet the needs of various customer classes he said. He further discussed how orchestration can handle servers which go down by spinning up new functions on other servers and gets you back up and running. Dynamic scaling allows a virtual network function or VNF on a CPU to be monitored. So lets say utilization gets above 60%... At this point you can spin up new instances to handle growing demand. Conversely, you can spin down servers as needed.
He advised that everyibe gets VMWare and Openstack running in a lab. He suggested using old servers for testing - this isnt producted he suggested. He says we need to learn new ways of operations management.
I have been in the industry for more than 20 years and I havent seen anything ad exciting or as earth-shatttering as NFV and SDN.
Day 2 Start
Rob High , CTO and IBM Fellow, Watson Solutions, IBM Software Group takes stage
Keynote title : Cognitive Computing - The Brain on a Chip
The session starts off with a video touting how advanced Watson is - the video seems oriented to developers who can access the company's APIs.
The discussion moved quickly to the challenges Watson had to overcome to be able to successfully performcognitive analysis. Much of the discussion so far centers around the game show Jeopardy where Watson successfully competed. A complex flowchart shows us the reasoning that Watson employs. He says they are looking for weak signals in noise to help them improve accuracy.
In closing, they think cognitive computing will be huge in the future. They have invested over a billion dollars into this area and they have an ecosystem allowing ISVs to develop applications and deliver them content. They also have a $100M fund to help launch new companies in this space.