IoT, 5G and LTE efficiency can all be aided by web scalability, according to Ian Hood, Red Hat's chief architect for global service providers.
Last time, in Part 3 of this Q&A series, Hood talked about the general considerations of web-scale operations and services -- and how open source, community partnerships and software-defined networking technologies play a part therein. (See Red Hat's Hood on Blending Open Source, SDN & Web-Scale IT.) (Earlier, in Parts 1 and 2, Hood focused on software-defined networking in the SD-WAN context. See Red Hat's Hood on Perpetuating SD-WAN Success and Red Hat's Hood Talks Security for SD-WAN.)
In this final installment, Hood gets into some of the practical realities of web scalability for network carriers.
Telco Transformation: There have been pricing and network inefficiencies with voice versus data with mobile networks and LTE. Is there some interplay to be found here with web scalability and EPC?
Ian Hood: There's sort of a synergy, and that was kind of the goal of LTE in the first place -- to make it so that I could do both voice and data and slice up my architecture [down] at the packet core such that voice calls did what they needed to do, and data calls did what they needed to do based on their setups. But that was a certain set of architectures in 3G.
For LTE, we put the data on the LTE side, and left the voice on 3G. Now with the advent of being able to do voice at the quality that we need to communicate over an LTE network, I can actually turn off the 3G network and just do everything in a data mode. So in effect, voice now is still a special flavor of data from a guarantee perspective, expectation-wise, for those who still make those calls because we still make, for the sake of argument, five to ten calls on our mobile phones on a daily basis, versus ten to 50 texts a day, versus how much time we spend just actually pushing data around the network. It is going to change to a common technology rather than two separate technologies, and that does give some advantage to the operators to use consistent means to deal with that application in a common way rather than two sets of protocols and two sets of networking gear, etc. And they're going to do that before 5G.
And so really all that 5G is going to do to that picture is take that consistency and make it so that I can scale up those consistent resources for whatever type of service I need to offer because now I have a different flavor. So we've got data, we've got voice --
wonderful. But now we've got sensors that send little tiny bits of data constantly or whatever, and now we've got trillions of sessions to deal with that I have to go collect and aggregate and decide which ones to ship across the network to the rest of the world. I've got to deal with another flavor, but at least I have a consistent way to represent that connection or collection of connections from a networking perspective. Then I have an easier job than having yet another bespoke network just for IoT.
Get real-world answers to virtualization challenges from industry leaders. Join us for the NFV & Carrier SDN event in Denver. Register now for this exclusive opportunity to learn from and network with industry experts --
communications service providers get in free!
TT: So what more can you tell me about how a web-scale IT environment can help enable IoT and Smart Cities?
IH: In that respect, the advent of IoT is for each vertical to improve that business and how it's run; that's its real purpose in life. How does a city perform? How can a city do that more efficiently and offer better services? The same thing applies to manufacturing. The same thing applies to healthcare. It's really about, "How do I improve the delivery of the end service?" And, as a result, if I can automate those processes for the technology through sensors, etc., then that's how I can actually improve that. As I go to automate those things, and I use devices that are now automated and controlled that I can spin up or spin down, it becomes that elastic nature of the architecture. So IoT is just a specialized flavor of that distributed architecture that needs to have its resources spun up and down to meet the improved service quality for a city, for a manufacturing plant, for oil and gas and for health.
The number of devices that I could potentially go drive adds yet another set of complexities to the scaling scenario. But with that IoT architecture you don't need every piece of data from every sensor to be sent everywhere on the planet. You have to actually have intelligent architecture that is smart enough to say, "Well, I need to send this information locally, this information goes upstream to the next place, etc." It becomes an architecture of sending the right messages to the right place over an open bus between the edges of the network and the data center where the actual end-IoT application might live. This web-scale approach is going to be applied in that same fashion to facilitate instead of being macro-network resources to being the resources associated with supporting the application for that manufacturing plant, that retail facility, that hospital or that city.
TT: Because IoT, as you scale out, can help measure these metrics and deliver and analyze data more efficiently, how else can the benefits of web-scale services and operations be relevantly quantified -- whether in terms of availability, efficiency or other available metrics?
IH: The next thing is now I've got this data from IoT or wherever it might be or from any service, so the next real challenge for us is how to take advantage of that data and make use of it to improve your data or improve your services or your business. So that's where, on the other end of the spectrum, you take a look at it and you say, "What can I do with analytics and algorithms in that regard? And how do I crunch this data to improve my business?"
You've got a massive amount of devices to deal with out there at the edge. Going up this collection of data, deciding what data to go collect, now I've got to go crunch that. I've got to actually bring in high-end computing, artificial intelligence, things of that nature to take advantage of "What can I do with the data to improve my service and, again, improve the efficiency?" Then you go apply policy to what you decided to do with that information you collect on the analytics side. Now we're back to what compute powers do we need and where to handle the crunching of all that data?
TT: Let's answer that rhetorical question, then. What kind of compute powers do you need to help facilitate a web-scale IT initiative?
IH: I would say that where we're apt to do the kind of crunching and do that kind of thing, you're going to head back into what I would call supercomputing. Except that instead of it being in a single location and architected in a proprietary fashion, it's now a distributed supercomputer in multiple datacenters spread along the countryside.
— Joe Stanganelli, Contributing Writer, Telco Transformation