Cloud Native Computing Foundation KubeCon + CloudNativeCon China 2019 (Shanghai), 5 Jul 2019

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Virtualization, Automation and DevOps – Why it’s Taking so long for Telcos? - Anuradha Udunuwara

Description

Join us for Kubernetes Forums Seoul, Sydney, Bengaluru and Delhi - learn more at kubecon.io

Don't miss KubeCon + CloudNativeCon 2020 events in Amsterdam March 30 - April 2, Shanghai July 28-30 and Boston November 17-20! Learn more at kubecon.io. The conference features presentations from developers and end users of Kubernetes, Prometheus, Envoy, and all of the other CNCF-hosted projects

Virtualization, Automation and DevOps – Why it’s Taking so long for Telcos? - Anuradha Udunuwara, Sri Lanka Telecom PLC

Telecom industry has been talking about SDN, NFV, Cloud, Virtualization or what I call all of them together as “softwarization”, for quite some time. But, except for few telcos around the world, most of them seems to be struggling to achieve what was expected – “automation”. There are many reasons for that – technical and non-technical. How to put virtualization, automation and DevOps in place and use the tools like SDN, NFV, Cloud to achieve the ultimate objective of Digital Transformation – to become a Digital Service Provider or Digital Lifestyle Service Provider or whoever you want to call? What were the original roadblocks? Did we address them? What else need to be addressed to shorten the transformation and start reaping the benefits?

https://sched.co/NrvI

A

Hello, hello: everyone thanks for joining this session, it's almost end of the day, but thank you very much for turning up what I'm going to discuss this evening is about virtualization, automation and DevOps, but specifically, why it is taking so long for telcos. Why is the delay now we've been discussing about many things yesterday and today, and hopefully tomorrow, some great things about open souls and how open Souls is actually helping, not only the telco industry but all the verticals different industries and as a whole, everything that we do today in day to day life.

A

We have a lot of support from open source, but there are few things specifically in telco domain that things are not moving the way that we want sometimes and the whole objective of me, sharing this one with you, even though we have around 25 people here is that for some of the problems and issues that I'm going to raise, we need to find solutions together. We need to collaborate as an ecosystem.

A

For example, the telcos I represent Sri Lanka Telecom, the telcos and the standard developing organizations, the open source community, the vendors everyone need to collaborate and find solutions. My name is Anuradha who do Nora? You can just call me Anu and I'm from Sri Lanka Telecom I will show you some things about my company and Sri Lanka.

A

So these are things that I'm going to discuss today with you for the next 35 minutes, start with Sri Lanka in Sri, Lanka Telecom, and what are the sort of transformations that we have been doing in the past in the telco domain and white is not moving fast compared to that in other industries compared to the IT compared to the computing white is not taking, or rather accelerating, the way things happen in the telco. What are the reasons?

A

We will look at the hardware mindsets that we have in telcos, some of the complexities that we observe in open source when it is being applied in telco domains. And what are the roles the vendors need to play and as an example nfe and menno complexity and some of the skills.

A

Then we look at the current status of telcos and also what are the things that we can do to accelerate the transformation.

A

So this is where I come from Sri Lanka. It is a small island south of India. Anybody been to Sri, Lanka yeah, very good. So it's a small island. We have 22 million population in Sri, Lanka 65 thousand square kilometers in size, but we have eight operators, three fixed line operators and five mobile operators to serve 22 million. So it's a very competitive environment.

A

In terms of telecoms. Our mobile penetration is 150 percent right, but our fixed line penetration is around 12 percent. Our broadband penetration is around 34 percent, so I don't think we can do much on mobile because it is already 150 percent. Then it's almost saturated, but we have a lot of opportunities in expanding broadband, specifically using either mobile or fixed, and these are some pictures. The reason I am posting. This here is actually Lonely.

A

Planet recognised, Sri Lanka as the number one destination to travel in 2019 and those who have traveled know exactly what I mean so I recommend and rather request others also to consider travelling to Sri Lanka nice place to be Sri. Lanka Telecom is the incumbent operator in Sri Lanka, like China Telecom, China Mobile. In China. We have a history of 162 years right, pretty much diverse into many areas. We have 11 subsidiary companies working on content working on several other things like this is our own university that we run for education. This is for healthcare.

A

This is for mobile and this is for digital publishing, managing human capital, managing properties and also we run our own submarine cable maintenance, Depot in South of India south of Sri Lanka. Our revenues come mainly from fixed, but we have a very growing demand for mobile as well.

A

So we recognize ourselves as the National ICT solutions provider and our objective is to become a digital lifestyle provider by 2022.

A

Alright, so my involvement at Sri Lanka Telecom today is I work as a senior engineer for enterprise solution designing. But this is my 16th year tree, Lanka Telecom I have been doing mostly network planning, planning, IP layer, 2, Packard, etc. Those networks, but since 2015 I am also engaged with our digital transformation, meaning that we are trying to transform ourselves as an organization, and we are trying to use software ization capabilities like Sdn, Software, Defined, Networking, NFV, Network, functions, virtualization and also cloud.

A

Therefore, I have been doing my own research and also I, engage with two universities, one local and one in unit United Kingdom working in these areas and as a result, I have been talking at most of the conference's happen around the globe. Since 2015 and one of the king's consistent thing that if you carefully look at I, always discuss when I get an opportunity is about the challenges telcos are facing now the reason I'm doing that is actually to pass the message.

A

Get people make enthusiastic about the problems, discuss the problems and find solutions collaboratively now. I expect from all of you the same thing like if you think that there are things that, as as an ecosystem that we get together and find solutions that will benefit not only you but also the telco industry as a whole.

A

Let me start with the business problem now. This business problem has been quite consistent for the last five to six years.

A

What we have seen in the telco domain is that we are moving from voice dominant business era to a data dominant business area with the inception of internet and packet networking, and what we see since then, is that our traffic, whether it is bits per second transported or so many bytes stored, is increasing I think you all agree with me now what happens as a business organization to build these capabilities, or rather to deliver these capabilities to the end customer, whether it is enterprise, mobile or residential, we need to invest lot of money in terms of network modernization in terms of increasing capacities, building new data, centers, etc.

A

On the other side, what we also see is that our revenues in general are not increasing, and also, if you take the mobile example, 150 percent penetration means we cannot increase the number of customers also now this results, this divergence or the decoupling between the network traffic and the revenues that we generate. Now.

A

This is a common scenario very well observed and very well accepted by the industry, and we, as an industry, tried to find solution to minimize the gap or rather bring the two curves together again now, if you look at mathematically, consider this as a mathematical, two graphs, one thing you can do is bring the cost curve down. The other one is increasing the revenue curve up pretty simple, but it's not easy to do in a telco business environment, and we thought of different things to address that now to bring the cost curve down.

A

Things which were proposed by the industry is true. Try automation, do virtualization in the data center and also in the network, and that is why NFP and Sdn, and try to do development operations, DevOps practices across the organization.

A

In that way, you can bring your cost down, because automation means it is less human intervention so that you can automate things. Virtualization means you have less resources, but you are optimally. Use those resources. Development operations also help to bring the cost curve down. On the other side, how do you increase revenue now?

A

One of the problems the telcos are struggling to increase revenue is because they don't have new products, new solutions. You cannot get new revenue without producing with now proposing new products to the customer. Now why it is difficult to come out with new products is because we lack agility and flexibility in terms of developing new products. Take an example: if you take Amazon Web Services, for example, now the way they come out with products, the timelines and so many days or so many weeks they come out with X number of products.

A

Take the same thing for a telco environment. Can we come out with, let's say five products within five months? It's not easy. I will explain why it is and because we lack agility and flexibility. We cannot come out with new products because we cannot come out with new products, it's difficult for us to increase our revenues.

A

Now, if we introduce agility and flexibility, we can, on the other hand, come out with new products and increase our revenues.

A

Now to address this problem, we have been trying different things, one of the very remarkable things that we did as an industry is actually moving away from the legacy circuit, switching time, division, multiplexing type of networking to packet, switching IP and Ethernet type of networking.

A

Now that resulted something called next generation network I hope all you have heard about next generation, Network and I'm, pretty sure that you have not heard about OGN OTN. It's a term that I have introduced stands for old generation network, there's nothing called generation Network, but for the perspective of understanding, what is ngn the one before? Let's call it oil generation network and what we are trying to do, because, since the problems are not solved by this transformation, we are looking at another transformation.

A

This transformation is quite different from the transformation that we have already had in the past, and the new network I used to call it as software generation network. It is nothing but the same virtualized version of the next generation network. The meaning is that we are not actually challenging any of the IP transport technologies or theories. We don't say that we don't use IP. We don't use TCP, not like that.

A

We will remain using the same transport or transmission protocols, IP Ethernet, MPLS, etc, but we're trying to were chill eyes the network, because virtualization in the IT world or the computing world has actually proved that it's the way to do things and we are trying to do the same thing in networking, and that brings us a new term called soft terrorization.

A

Now the place that you can actually find the meaning of software isolation is from the I Triple E software as asian study group, which they say is that it's a combination of Sdn, NFV and cloud so by using a sneer, nfe and cloud we're trying to come out with a virtualized version of mgn and as a result, we are trying to move away from network mindset as a telco to a more like a platform mindset where these cloud companies, web-scale companies and over-the-top companies, I actually started with.

A

So the reason that we have to do that is actually we want to move from what we are today: the communication service provider to one that actually provides digital services to end customers or a digital service Freud. So the business objective for us is actually moving away from communications as a service, but to actually offer digital as a service. So this is the business objective. How do you realize that business objective is actually by transforming this mindset? It's not a net technology transformation.

A

It's the kind of mindset, transformation that you have to do from Network thinking to platform thinking. It's not a network that we have it's a platform we have so what's the platform, it's a network plus the data. Centers plots everything it's more like the web scale, IT thinking and what helps us to achieve that is software as a sh and Sdn nfe and cloud are the tools right. Ultimately, we achieve the business objective now why T have been so difficult for us.

A

If you take my company, it is 162 years old when a company is that old compared to companies that 5 to 6 years old, we have new startups here now as telcos, not only us any telco organization. If you talk to, you will understand that we have very distinct mindset, called the hardware mindset, because it is okay, because we started with hardware. We have started with Hardware switchers. We had started with Hardware routers. We still have hardware, we think a network is set of hardware connected with wires.

A

So that is the mindset that we have as telcos and on the other side, when we want to deliver a service, whether it is a residential business or enterprise customer. We think that again, we need a hardware at the customer premises and that need to be connected to the network, which is also a set of hardware, and that hardware we usually call as a box.

A

So we have this mentality that we always need a box, a box which is visible, whether it is at customer premises or in our network or in the data center, that we need to see that we, we are very proud that if the boxes were a big right, we have routers of this height. So the other thing is that when you have the box, we also think we tend to think that we need to configure this box. So we have this mentality called box and configuration on the first place.

A

We think that we need boxes and the second place. We think that these boxes, whether it is it's with ourselves or with the customer premises, we need to configure that so this box and configuration mindset has been evolved for 160 plus years for us now. This is something that it's very difficult to. It's quite rooted to our mindset. I'm not saying the people which I work with the 160 years old, but as a thinking process.

A

What we have been inherited over the years is that this thinking process, the network systems are boxes, boxes need to be configured. This is what I call, as the hardware mindset now take an example, a classic example of Sdn or software-defined network, something which was started at Stanford in 2005 around that dear. Let's call that as Stanford sta there's nothing called Stanford steel, but for comparison sake, let's call it Stanford this deal. We were quite successfully to Kesley into the data center. We also have a data center running with this da. So it's pretty easy.

A

Then we try to take Sdn to the packet layer specifically for IP, not that easy. But there are a few few operators who have already done that.

A

Then we are talking about taking as the end even below 2.5 layer, 2, layer, 1 and also taking even up to the photonic layer and try to not really deal with IP packets as such, but rather deal with the wavelength in the optical domain, using software-defined networking, because when you say software-defined networking as a concept, it does not restrict you whether you should apply it for packets, friends or even wavelengths. You can apply 20, but if you really look at this level, it is more an academic discussion than an implementation.

A

Why I am saying that if you, those who are familiar with the telco terms, if you look at the technologies that we have been introduced in the past now we have technologies like ADSL, MPLS ISDN. Those technologies were not restricted to one particular geography, something which was introduced in United States, let's say MPLS at same time. It was applied in Asia, whether it is a developing country or developed country.

A

There was some uniformity when the technologies were introduced to the telco domain, but this time, when you take technologies like Sdn, nfe and cloud, we don't see that uniformity. We see only few Tier one operators actually having few success stories to share on their success. But if you look at the companies coming from emerging economies and developing economies, we don't see that success stories with them now, there's something that we need to look at very firmly white is so.

A

Now take open-source, for example, open source, as you all agree is, is it's a very novel concept to telecoms and also to the entire industry, and this is the slide I presented in 2015 in Barcelona at that time. The reason I wanted to use this slide. I wanted to discuss something on the number of options that we have. For example.

A

These are the options that you have to implement and Sdn controller. This right we have, then we have cloud management. Few options so I was trying to discuss about having options are good, but we have to select among options which one is the best and which is to be used now, since then, I have been updating this particular slide. What I observed is that, specifically in not necessarily in open source domain but also standard developing organizations, we see sometimes the ideas collaborate. Sometimes the ideas diverge again.

A

Collaborate again diverge right, which is one may say it is good. One may say it is bad, but there are certain implications happening specifically in this domain, so within a period of four years, I find the slide becoming much complicated. Now we have few options, for example,.

A

A

First, we so open networking foundation collaborating with owen lab becoming the new onf, and after that we saw many projects which are related to onf like this, and then we saw for some good reasons. Most of these projects were actually hosted within the linux foundation.

A

Example open, daylight, open, V, switch.

A

Panda, which is platform for network data, analytics a very important project and also owners and also own app, which is a combination of something actually started with AT&T and also open o combining that we have ona, and then we have code. We have p4 as a language. We have open platform for NFV and Etsy defining NFV and we have another option for orchestration, which is open, source manual company compared to own app. And then we have another option for Sdn, which is tungsten fabric like this, and we see most of these.

A

Orchestrators are using OpenStack as cloud management and very recently we saw seven of the projects which were under the Linux Foundation, which were hosted within the Linux Foundation. They sort of collaborate at all that harmonized to make Linux Foundation networking something similar to Linux Foundation age and also Linux Foundation, deep learning, but not quite so. If you really look at the Knux foundation edge, it's it has overall blueprint for that, but we don't see such blueprint for Linux Foundation networking, but we have some hope that we have few options. Selected options collaborated within one umbrella.

A

Now this is quite complicated. I'm, not saying that these projects are complicated. If you take individual projects, they're actually doing really great thing. But if you, if you look at the same thing, if you face this as a telco now, if I am asked to select an SDN controller and if I am asked to select an open-source or not Orchestrator cloud management. Software now I am given so many options to select now as an industry.

A

Having options on one side is good. But when you have so many options and when you are debating on which one is good, which one is bad, that it's not helping the industry to move forward, and that is one of the problems that we currently have, which I think which is actually degrading. The progress of telcos, moving into open source and also transforming themselves.

A

Now, one of the reasons things that compared to orchestration onap versus open source manner, now, if you take this, is all information which is available in public domain. If you take telefónica, which was initially wanted to do, orchestration with onap recently decided that no, we are going to use open source manner. Now there are Tier one operator now. If the Tier one operator certainly changes directions like that.

A

Imagine an operator from an emerging economy, small operator will it be easy for them to select which one is best, which one is bad right, so we're actually trying to use both in our environment, because we cannot really select or rather go and tell vendors that you should supply or nap or you should supply OSM say that you should supply both now coming back to winters now in open source.

A

When I talk to most of the people who are demonstrating things, they said that everything is open, east-west north-south, all the AP I saw open, but many when it comes to tell codomain when you're talking about hardware integration, routers, switches boxes.

A

This is a question that we have. Are they truly be willing to open right?

A

The true openness for me is that you have, when the a and when the B today integrated using an API or an interface tomorrow, you want to replace when the B with vendor C, so you remove in the be bringing vendor, see if you can do that without doing any modification to win the a side that is true open source, not open source, true openness, but we don't see that openness today in the industry.

A

So when the vendors say that we are truly open, we have to actually go and test, and then we find what is the openness now, if you take network functions specifically in telco domain, we used to two things with physical network functions, the boxes again, and we think that, even though this is quite complex, we are able to manage it today and tomorrow. We are thinking that, with transferring all the physical network functions to virtual network functions and it should be simple.

A

That is what we think, but the problem is, you cannot just go from here to here within 24 hours. You have a transition period where you have to deal with vnfs and PNM's. The intermediate part that is the most complicated thing and orchestrating physical network functions and virtual networks function in a migration situation, is not easy in a telco environment, so orchestration is challenging and I will show you an example. If you take.

A

Because you can have your applications, your virtual network functions hosted in many places. It could be in a bare-metal, it could be in a container, it could be in a virtual machines, or virtual machine could be inside a container or concered container could be inside virtual machines. There are many ways that you can host an application today, but when you want to, for example, do a service chain, a complex service that you have offered to enterprise customers, which is actually chaining in between these particular servers.

A

You could easily have something like this for one particular service. Now. The thing is: if the customer complaints you, that my service is not working, how we are going to troubleshoot this in a telco environment right because you don't see any of these things, you have to imagine that this particular service is there and it is connected to that and where I'm going to troubleshoot it and how I'm going to troubleshoot it.

A

It's not easy, and for most of the problems we see that the skills that we are having today the knowledge, education and the skills is the major barrier. The most the most difficult thing to address is this one, because this is from telly management forum, 2018, sort of 2017 findings. 80% of the CSP is thing. Lack of employee skills is a barrier for transformation and 70% thing that retaining skilled employees will be a challenge. So on one side we lack skills. On the other side, we find it difficult to keep the skill people within right.

A

So when you don't have people who actually have the right education right, skills and right knowledge, it's very difficult for us to transform ourselves now. What resulted because of all these is that you have standards moving. If you stay in this conference till tomorrow, you will understand that standards are moving very fast. We come out with new terms new acronyms new standards, new open-source projects, which is good. On the other side, we see vendors with all due respect.

A

They also implement these standards when a standard comes out open-source community, it is also available through a vendor as a as a distribution or a release, and also, if you take our end, customers the end customers of telcos, they also transform themselves pretty fast. If you take the banking industry that is in my country, they are the number one transform and Enterprise in in sri lanka now what happens to operators and service providers in between these three, we are kind of sandwiched.

A

We are struggling to move forward and, as a result, in this particular race, different people travel at different speeds. Now, when you travel at this speed, it is very difficult to take the whole industry port because it's no point the standards moving fast vendors moving fast. But if the operators are lagging, then we are not moving forward as an industry. So it's very important that we help operators collaborate, discuss and help them so that they also come to the same pace as we move forward in the total transformation.

A

What are some of things that I can propose, or rather, as an industry we can propose, is most of the things we have to do ourselves, because our problems are our problems, because the local problems cannot be addressed by somebody outside now. For example, we need to set clear and aligned missions for ourselves as telcos. We need to change the organizational culture as I told you, the culture that is evolving for 162, plus years with hardware and network mindset.

A

We need to change that to software and platform mindsets, it's not easy, then you need to drive the transformation from top to bottom. It's very difficult to do it other way around. You need to get the leadership driving that and we also need to understand the real ROI behaviors of software isolation, investments you need to when you want to get your budgets approved from the CFO. You need to explain him that you don't get from day one your cost going down. First, it will go up, then it will go down.

A

So when you are getting your budgets approved that you need to actually convince them. This is the real ROI behavior of soft terrorization investments. Last but not least, the most important thing is identifying and bridging the skills gap. If you have skillful people knowledgeable and educated people within your team, it's not easy. I mean it's easy for you to move forward. So in summary, sub terrorization of networks and system is a key for telco digital transformation.

A

However, it's difficult to do that with the current hardware mindset. There are many challenges, mostly non-technical, and they need to be solved to operators themselves and skills will be the deciding factor when we move forward as an industry, and these are some of the reference. This is available in the slides, some of the blog posts that I have posted on the same same topic. Thank you very much.

A

Any questions we do, we have time for questions.

A

B

Many side roads over there and new technologies right, but but rotation comes to the market we have to. We cannot treat everything to the new technology overnight, so we have to live with the heterogeneous environment, and some services like like mixed wine, have to be kept.

B

It is specified by row, so we cannot arrange a kind of services, so at least in Japan. Some old boxes have to be the main without migrating. So and, as you said, I understand that those all the machines box is quite old, like 40 years old over 50 years old, and it is very hard to maintain and the total amount of people who can maintain those equipment is limited and they can they can manipulate old machines, but cannot understand new technologies right politician. That is a problem.

B

Likewise, I'm, not young, but I cannot understand the old equipment. However, I understand somehow new technologies I understand that there is a gap and it is very hard to overcome. I think it is very important point, but I mean the early. Sometimes it is very hard to overcome. Let's.

A

B

Agree with you and one question to ask I understand this meeting is focusing on container container night's applications and do you think that's technology like containers helps to make station.

A

It's for me, it's not I, think we are running out of time. It's not not really the containers, but how do you actually sort of package your services I, would call it micro services, so it's it's, it's just the size and how flexibility that you are getting from these things. So maybe we can discuss further I'm signal that time is out. Thank you very much.