MobiledgeX, Wow! Listening to Jason Hoffman, CEO
These are highlights from Jason Hoffman (JH), CEO of MobiledgeX video interviewed by Iain Morris (IM), News Editor at Light Reading
|Jason Hoffman (right) and Iain Morris|
An entire new different companyIM
You just joined an entirely new company, MobiledgeX. Can you just give us a bit of a background? What it is exactly what you are doing there? What prompted the establishment of this firm?
We are technically a spin-off of Deutsche Telekom. We are a wholly owned subsidiary that is focused on edge computing, but in particular our focus is on developer facing edge services. These services help new developers on new devices, or new use cases of existing devices as well as required infrastructure transformations needed to deploy edge
What prompted the move? I mean edge computing has been around for a while and Deutsche Telecom was looking at these and other operators. It is quite an unusual thing for an operator to create a spin-off startup which is wholly owned rather than doing the work internally.
A Deutsche Telekom spin-offJH
I think Deutsche Telekom has a pretty good history in the sense of (spin-offs). Each country is sort of naturally a spin-off or a separate company under a larger umbrella, the number is, I think, in terms of high tens of companies being within that group. So for me one thing that was attractive is that it wasn’t new for them. Very often when you want to do a new effort and really have a focus on it, it is a good idea to do it as a stand alone
As far the timing around edge computing, the reality is we had two parts of the infrastructure undergo a pretty dramatic industrialization over the last ten years on. On one side and the telephone space IT was largely run. In the case of LTE - we know how LTE had to happen to support things like iPhone and so on. You had things that really fit almost hyperscale definitions of a very distributed infrastructure.
What MobiledgeX does?You can see the complete opposite end of the infrastructure, where you had all the hyperscale cloud providers emerging right now.
The reality is when you look at this infrastructure that is between those two, that also has to undergo a pretty dramatic industrialization effort
We are starting this effort from a developer standpoint and say: “Well, there is the Internet. Three’s mobile networks and mobile networks are not the Internet”. All these capabilities exist you know and normal sort of Internet world that don’t actually exist entirely on the normal networking side, but if they did then there is a lot of interesting things we could do
If we look not just at the devices that exist today and interesting applications you can do for them, but all the devices that don’t exist (yet!) because either the economics of the device or the capability of mobile networks. Look at the emergence of the sort of Augmented Reality (AR) type use cases, Virtual Reality (VR) type use cases, always-on type use cases, isn’t the like you know. The truth is everything basically in the mobile networking world got to have a full set of capabilities to really do those applications
We are essentially talking about moving those IT resources out much further to the edge of the network closer to the end user which is an investment that operators have to make. Where does MobiledgeX fits in exactly?
Where and what the edge computing is for you?
For some reason, there is almost an over obsession of where and what their edge computing is right. Yes, but what I mean, the truth is when you look at the collection of smart phones and platforms in the world, it adds up to a lot of CPUs. You look at a bunch of infrastructure that in between it adds up and you look at public clouds. I don’t thing turning into another and workload sort of moving you out
If you had things like anybody that’s doing an application on a smartphone today, and you do location services on the smartphone, it is what dramatically drains the battery life. The only reason is that we are doing locations on the smartphone, is because there not an edge service available to you, that tells you very precise and accurate location. This not the way location services have been done in a telecom before, but like a uniform way of doing that.
Only when you look at very simple things that can be pursued to doubt the developers, you know in the case of mobile networks, these can have dramatic impacts on the battery, on the performance of the device and so on. I think a lot o those operational services we are at the right time to do that.
MobiledgeX challenges spelled out
What are the Big sort of challenges in the edge computing area that you specifically looking to address?
Well the Telecom space we were traditionally not as good as it could be for developer facing things. One reason we wanted to do this is as a separate company is in given markets all the operators are going to work together. They present to give a view to doing that to a stand-alone company. This is easier than doing it as part of the competitors in the market
We have sat there working with a lot of current mobile gaming developers – they are AR and VR, types A lot of people working on new devices like security cameras. We have a set of pretty common requests from them that said; “It will be really great if you know I could get this feature or this type of information” I can do that and we are trying to be that company that is taking a lot of the great aspects of mobile networks, actually making them transparent to developers. We are creating those (request fulfillments) by focusing on the services
We already started
I think Deutsche Telecom was talking about bringing other operators in as well. How about having those involved in mobile objects. Is that something already started to happen?
Operators work very well together when are working, I guess, indirectly? (Jason smiles). There plenty of historical examples for that. You have to help the whole community and execute well at the edge.
5G helps the edge
(Miha's note: with so much hype - like 5G will save humanity - here are specific intelligent visions involving latency)
How important is in the context of 5G? I mean low latency services?
You have a really simple thing. For every 5G vendor the whole architecture of 5G is disaggregated hardware and software and every vendor for example is standardized, but really each is a set of microservices running on Kubernetes. Every vendor in the space has picked a distributed container platform. To deploy literally all the 5G software in it. the transition to 5G means that you know the millions of base stations in the world are going to be talking to Kubernetes clusters. So, what else should we be doing? For example, the idea that all we’re doing is running radio control software. It’s probably not quite right. So, when you stop thinking about that, there are going to be tens to hundreds of millions of small cells talking to tens of millions of microcells, in turn talking to hundreds of thousands of aggregations sites, and those talking thousands of regional data centers talking to thousands of national data centers? In that perspective the whole sort of like topology that’s in there and then even at the very further edge if we are trusting container platforms like Kubernetes to run 5G control systems? OK, then we have some really interesting deployable spots inside of Telecom to put new ideas
How quickly do you see we are taking off with talk about 5G? This is one big themes of this show and its really what is going to give the push to edge computing in some of the use cases that you are looking now.
Other big conversation that we have here is the business case for 5G. I think this business case requires edge
For (devices) that require low latency…
Not necessarily low latency, but no latency. If you are looking to an autonomous car going up the road, there no way that any operator can guarantee. Say 1 millisecond continuous latency on the whole road. It doesn’t work that way. What we could do is actually expose known latencies to people so you know where the base stations are and what the topologies are when you are at this point of the road where you probably will at 28 milliseconds so you going down to 4 milliseconds and then back up to 8 ms that you are going to hand over again
These developers there don’t even know that information to even being able to do that type of work on the car
What mean is that low latency czar you know 5G just sitting there and saying (the access latency czar) you can get down to 1 ms, and you say OK! We get it down to 1 ms latency and that is a lot of things that would be interesting to do if the entire cycle of doing it could be less than 100 milliseconds in total. Then the access time is probably not going to be the limiting factor anymore. Somebody like NAQDAQ tried to do a trading in about 22 to 20 microseconds Inside healthy data center high hundreds of nanoseconds and the complete cycle might be 1,000 milliseconds’. Our connectivity times in a sort of pre-5G world have been the rate limiting thing. Now in 5G they are not going to be rate limiting thing.
We will see what happens. Latency is only one part of it. There are other big part you actually have those latencies known. Then you can reliably offload computer and data from the device. You can then start having devices where you are not doing battery intensive location services, wecan do them with like new 5G location services. You are able to sit down, and do something like a virtual desktop delivery.
We can allow devices to either be a longer life from a battery perspective, of make smaller form factor that is cheaper. If you can take a 3,000 dollars headset and make it a 300 dollars, you know this has a big impact
When did you see some of these applications appearing in the trial stage perhaps?
MobiledgeX reminds about how AWS started?
We have been doing trials with developers already in Germany. I mean we have a couple of locations up and running, The intention is to create in Q3 and Q4 this year we continue to come out with new ones. A lot of these are simple and powerful. Remember that AWS (Amazon Web Services) started with a developer payment API, then to a an API called messaging cue system and then a three API called optic store and then you know a compute service that was really just batch compute on the object store. The documentation for the whole thing was a couple of pages.
There is a lot of things we have when you talk to developers interested in these. They say: “look if I could just know these two things that influences this one thing” – meaning if I could do to get requests on an API and a post, I’d be pretty happy and this has this much value. If there are a bunch of people like that, we just want to launch those we know. Sometimes we don’t know them. But we are trying to get on this type of cadence where you know if we come our with something like that, launch it.
Thanks very much Jason
Thanks for having me
The interview is part of MWC 2018. In a News Analysis published by Iain a week later, he quotes Jean-Claude Geha, a high level Deutsche Telecom executive
"The ideal partner, it seems, is one that has fully embraced the "cloud technology mindset" and the collaborative working style that goes with it. Deutsche Telekom wants "real cloud technology" as opposed to "slightly modified platforms." It expects some degree of knowledge transfer, too. "We don't have to know everything but we need to be able to work end-to-end with our suppliers and third-party integrators,"Jason is the ideal brain for this Cloud focus in telecom. He is ex Ericsson, and so is Doha.