As enormous legacy organisations with long operational histories, telcos sometimes lag behind other sectors in digital transformation – but cloud native offer a chance to jump a few phases. What challenges will telcos face when moving to cloud native? What role is Kubernetes playing? And in what meaning of the word should open source software be ‘free’?
Ahead of his presentation at the Layer123 World Congress 2020, Anuradha Udunuwara, Senior Engineer at Sri Lanka Telecom, discusses the lie of the land when it comes to telco cloud native.
Anuradha, what are the benefits of cloud native transformation for telco companies?
Let’s start with a little bit of history to get the context correct. Rather than investigating the benefits, telco digital transformation (in the business sense) and telco softwarization (in the technological sense) have now almost become inevitable for telcos. The requirement of becoming a digital service provider or a digital lifestyle service provider in the true sense, making radical changes in the way internal (operations – service delivery and maintenance) and external (all customer interactions) processes are run, is not doubted by telcos as they clearly understand the consequences of NOT changing/transforming. However, telcos being traditionally massive in multiple angles – massive networks, massive workforces, massive subscriber bases etc., have been quite reluctant to change until recently. This is mainly due to the culture they have inherited for the last 100+ years – 150+ years for some. No-one is to be blamed for this, including the telcos themselves, as the hardware centric networks and associated hardware mindset of the workforce, with widening skill and knowledge gaps, have become one of the biggest showstoppers for CSPs becoming software centric and obviously having people with a software mindset. Again, the importance of being software centric in today’s business, not only for telcos but for virtually any organization, is well understood by all stakeholders of the ecosystem, as the agility and flexibility that “software” inherently brings to any business operation is necessary to survive in today’s ever changing hyper-connected digital economy.
Now, in the race of innovation, change, transformation or whatever the term you use, telcos have been, in general, real laggards compared to the vendors, SDOs (Standard Developing Organizations) and end customers (especially the enterprise customers of telcos). The pace at which new technologies and capabilities are developed by SDOs and especially the open source community/organizations, and the pace at which the vendors make available those capabilities in their products and solutions, is quite fast compared to the pace at which the telcos traditionally adopt those capabilities. This is again due to the points discussed above. The rapid innovation which made the telco networks transform from NGNs (Next Generation Networks) for example, to SGN (Software Generation Networks) with underlying capabilities like SDN, NFV and Cloud (altogether called softwarization) has been almost given a surprise by the new kid on the block – Cloud Native. Now, instead of virtualization, we talk about cloudification. NFV becomes NFC (Network Functions Cloudification). VMs become containers. VNFs become CNFs, and so on.
But, what is Cloud Native really? If you ask 10 people, you get 10 different answers. For me, Cloud Native is nothing but taking the full advantage of the unique capabilities of the cloud. As we all know, cloud capabilities include measured, on-demand self-services, broad access methods, resource pooling, elasticity and open APIs.
If you take the cloud and Cloud Native adoption of telcos, we see that on one side, the telcos that have already embarked on the softwarization journey will have to rapidly adapt to the Cloud Native environment. And on the other side, the Telcos that have not yet started their softwarization journey now have a golden opportunity to leapfrog to Cloud Native, very much like moving from Frame-Relay to MPLS, bypassing ATM. Anyway, for me, Cloud Native Transformation is an extension of the same Softwarization. The capabilities we originally wanted to achieve like automation, virtualization (now containerization!) and devops/devsecops work practices remain unchanged. Therefore, on benefit wise, Cloud Native Transformation, if we call it that, will deliver the same benefits of softwarization such as bringing in the flexibility and agility to introduce new services with short time to market, making telcos earn new and improved revenues at the same time as reducing capital and operational costs due to automation and virtualization, for example. At the same time, Cloud Native will for the first time give telcos the opportunity to become true cloud players or hyper scalers, as the very nature of Cloud Native really supports openness, micro services and massive scalability and avoids vendor lock-in, provided that the telcos have the right wish, will and skill.
What would you say are the main challenges to transformation in the Cloud Native model for telcos?
As explained above, the main challenges will continue to be culture and skills. Telcos, including their staff, need to transform from hardware mind-sets to software mind-sets to and now, Cloud Native mind-sets. That’s not easy. Skills, knowledge and education plays a very important role here. The other challenges include lack of overall company vision/alignment, including that of the top leadership towards Cloud Native business, and not understanding the ROI of softwarization/cloudification.
As telcos step in to a “cloudified software world”, fitting the existing legacy software into a Cloud Native world would also be challenging. Just migrating the existing software into a VM or to a container or just hosting it in the cloud – public, private, hybrid or multi, would not make the software Cloud Native. Some software/applications may have to be re-written from the scratch. For this you need a different skill set (for example, coders/programmers). The question is, who’s going to do that - DIY or not? If the telco does not have the in-house capabilities (or have at least not started building the in-house capabilities), then they have no choice but to go to the vendors and system integrators (SI), where the capability comes at a price. When that happens, and when you go back and see your capex/opex reductions goals, you find that you are not going to achieve them. That said, the best approach would be to start with a vendor/SI, but at the same time develop the missing skills and capabilities, so that you’d not have to depend on vendors/SIs indefinitely for everything.
Do you believe Kubernetes will drive Cloud Native for telcos?
I think so. Kubernetes (k8s) has already made massive strides into telcos. Started as a project within Google (Brog and then becoming Project 7) few years back, k8s came in to existence as an open source project around 2014. Kubernetes, meaning the helmsman or the captain of a ship in Greek, is surely doing justice to the name, by sailing multiple industries, not only telcos, towards the right direction. Being the top project within CNCF, the unanimous support that Kubernetes is getting from all the stakeholders from the telco ecosystem is quite amazing. This was quite apparent to me when I was invited to speak at KubeCon + CloudNativeCon and Open Source Summit China 2019 in Shanghai last year (https://youtu.be/J6whGFkQrqs ). I also ran a few surveys within the Linkedin and Twitter communities related to k8s and the results ascertained the same. For example, when the question ‘what is your preferred container orchestration engine?’ was asked, k8s got an overwhelming majority of votes compared to Docker Swarm, Apache Mesos, and others. According to a recent 451 Research study, 76% of enterprises will standardize on k8s within 3 years.
However, given that (1) k8s is relatively young and changing extremely quickly, now onto version 1.19 (from 1.0 in 2015), (2) portability, flexibility, control (ex:- full access to control plane), and in-house technical expertise are important factors to consider, many chose managed k8s cloud service over packaged k8s distro or open source DIY installation, when asked which flavour they would choose to build their k8s cluster. And today, for managed k8s cloud services, we have many options, including but not limited to EKS, AKS, and GKE. They also allow you to have on-premise versions like AWS Outposts, Google Anthos and AKS Engine on Azure Stack. When asked what infrastructure would they choose to deploy an on-premises Kubernetes cluster, many preferred a hybrid environment over virtualized and bare-metal.
Finally, with so many vendors in the market to help telcos (or themselves?), even for open source k8s, it’s up to telcos to decide how they want their open source software - free as in beer or free as in puppy?
Anuradha Udunurawa will speak at the online-only Layer123 World Congress - giving a talk entitled From Hardware Native to Cloud Native : No more an option for Telcos”. Registration to listen to this and hundreds of other presentations is totally free for network owner representatives and £200 for service providers - register today here.