Virtualization

Microservices: Applications, future OSS and anarchy

What the experiences of Comcast, Verizon, Twitter and Amazon can teach us about the move from monolithic applications to deploying microservices. Hold onto your hat.

This is an extract from our new, free research report, Microservices: Piecing together a strategy, written by TM Forum’s Chief Analyst, Mark Newman…register on our website and download it now…

Integrating NFV with existing (and future) operational and business support systems (OSS/BSS) is a challenge for CSPs. Indeed, NFV requires a complete rethink of network operations.

For instance, “We need OSS that is highly componentized, a service-orientated architecture,” says Barry Graham, Senior Director, Agile Business & IT, TM Forum. “Let’s take assurance for example: Before, you had a single platform for collecting, analyzing and reporting on data. Microservices now breaks this down. In a microservices architecture, you take the approach that data collection is just data collection. It offers componentization of functionality and processes.”

The other benefit is scalability, particularly when it comes to realizing the 5G vision. Long-term evolution (LTE) networks are not architected to be able to ramp up and take down different network elements. CSPs today “are virtualizing things that weren’t meant to be virtualized,” according Graham, who adds that today’s OSS simply cannot scale.

The Comcast approach – DevOps

Operators in the content and TV business, a market where OTT players are innovating aggressively, need ‘future’ OSS now. Consider US cable operator Comcast, one of the world’s biggest media companies, whose Enterprise Services Platform (ESP) powers back-office functions including customer and sales portals, interactive voice response (IVR) and Comcast.com.

In 2014 Comcast started transforming ESP with four goals in mind:

  • reducing time to market for new services
  • developing elastic scalability
  • increasing resiliency
  • decomposing the monolith.

Comcast started migrating the first services in 2015, but it wasn’t until 2016 when DevOps practices were introduced across the business that culture started to shift. The company plans to start using microservices this year. “We want to introduce the concept of a microservice doing one thing and doing one thing really well,” Christopher Tretina, Director of Common Infrastructure Platforms at Comcast, told delegates at the SpringOne Platform conference last July (watch the video below).

In fact, microservices are already driving some of the newer capabilities and applications Comcast is offering, including ‘Last 9’, a remote-control feature that allows viewers to see the last nine videos they’ve watched, whether on demand, on the web, a linear show or something on their DVR.

Microservices for digital experiences

Although OSS needs a radical overhaul, current thinking around BSS is more evolutionary. This is because there are some major obstacles to moving BSS into the cloud, such as national regulations around the storage of personal data and the need for charging and policy functions to be located close to the network to minimize latency.

Nevertheless, some leading BSS suppliers are embracing microservices for functions where privacy and latency aren’t a problem. Amdocs, for example, is adopting microservices for omnichannel and customer engagement systems.

“In back-end systems where you are less likely to change the code in high cadence, microservices are – at least currently – not entirely warranted,” says Zeev Likwornik, Head of the Amdocs Cloud Center of Excellence. “But for services where there is a higher cadence of code change – for example, a shopping cart in a digital store – microservices can significantly boost business agility.”

Large CSPs adopting first

Microservices are an extreme architectural concept, which can be used in cloud and non-cloud scenarios. However, for CSPs they are most relevant to cloud-native applications and represent an important part of the digital transformation journey.

It is no coincidence that the service providers deploying microservices are large, established operators intent on becoming innovative and agile, and delivering customer experience that matches web-scale companies. They have built large teams of software developers and engineers working across IT, network and enterprise functions. The expertise they have acquired has resulted in moving beyond virtualization to embracing cloud-native approaches.

In his keynote presentation (see video below) at TM Forum Live! in May, Shankar Arumugavelu, Senior Vice President and CIO, Verizon, referenced the importance of microservices to the company’s digital transformation. Microservices will be used to modernize the IT stack to reduce complexity – a process that has already resulted in the nationalization of two thirds of its applications.

But it is worth repeating that CSPs like Verizon are still at the early stages of this transformation. Greenfield systems and approaches will offer an easier pathway to cloud and microservices, but in practice legacy IT and networks will dominate the telecoms horizon for the foreseeable future.

Addressing the challenges

CSPs face many major challenges in deploying microservices, a task that is difficult even for large, established web-scale companies with considerable software expertise. Building applications as collections of loosely coupled services that operate autonomously requires a completely different approach to software development compared with building large monolithic applications.

Alex Roetter, former Head of Engineering, Twitter, once described it as moving from a centrally planned economy to capitalism: It’s difficult and can result in anarchy. When Amazon moved to DevOps and microservices, it found that the central engineering team which had previously been responsible for deploying new applications to customers was unable to service all these new microservices teams. So, the company created a tools group and a self-service approach to replace the central team.

More generally, moving from distributed monoliths to true microservices can be challenging due to stodgy culture, lack of appetite for innovation, lack of the technical skills required and not fully understanding the concepts.

Componentization is inevitable because it results in better products arising from specialization, focus and accountability. But in the same way that componentization has flourished in car manufacturing – because there is a relatively small number of global automotive original equipment manufacturers that can guarantee strong, predictable demand for car parts – so will microservice architectures only flourish in telecoms where there is scale, expertise and a willingness to learn through trial and error.

Download the whole, free Quick Insights research report, Microservices: Piecing together a strategy, now.



Advertisement:
Share.

About The Author

Snr Director, Research & Media

Annie Turner has been researching and writing about the communications industry since the 1980s, editing magazines dedicated to the subject including titles published by Thomson International and The Economist Group. She has contributed articles to many publications, including national and international newspapers such as the Financial Times and International Herald Tribune, and a multitude of business-to-business titles. She joined the TM Forum in 2010 and is responsible for overseeing the content of the Research and Publications portfolio.

1 Comment

  1. Terry Clarke on

    Give us a break OSS, NFV, BSS, CSP, OTT and not a single definition.

    OSS – Office of the Secret Service
    NFV – No Freakn Value
    BSS – BullS**t Squared

    Let’s see a return one to the fundamental rules of authorship – always define your abbreviations the first time you use it

    TM – Temporal Manipulation?

Leave A Reply

Back to top