AZURE KUBERNETES SERVICE(AKS): CASE STUDY

AZURE KUBERNETES SERVICE(AKS): CASE STUDY

SoftBank Corp. quickly develops an app with Azure Kubernetes Service (AKS), improves in-store service for customers

No alt text provided for this image

The IT Division at SoftBank Corp. plans develops, and operates systems in the domestic telecom provider arena (mobile and landline networks). The IT Division constantly seeks to advance its engineers’ skills and approaches to problem-solving and proactively create learning opportunities in new technologies. Starting around 2017, the division introduced the agile development method, wanting to take advantage of its ability to give young engineers lots of practical experience. Now, a new success story is being written at SoftBank using advanced technology to improve engineers’ skills. A business department supervisor serves as “product owner,” and the IT Division coordinates closely with the business department in a process of agile development. An example of that success story is the development of the Smartphone Adviser Reservation Tool (SMART), an application that supports the smartphone advisers who work in SoftBank shops throughout Japan. To develop SMART, SoftBank has adopted Microsoft Azure Kubernetes Service (AKS), a fully managed service for the open-source container orchestration system Kubernetes.

More smartphone advisers meant SoftBank Corp. needed a more efficient system

SoftBank Corp. shops around Japan have been full of customers looking for new smartphone contracts and the latest models. The company has made “smartphone advisers” a stronger presence in its shops in recent years.

Smartphone advisers are shop staff who specialize in support. They’re certified by SoftBank and use those qualifications to run smartphone classes and teach people how to operate smartphones. Since advisers were introduced in July 2015, their numbers have grown. There are about 1,200 employed across Japan as of July 2020. They’re very popular, and more than 2 million customers (often seniors) have worked with advisers to learn how to operate their smartphones or learn about plans that can save them money. In a SoftBank survey conducted in August 2019, a full 96.8 percent of respondents said they would use these services again.

As SoftBank’s customer-facing employees, smartphone advisers play a very valuable role. However, Tsuyoshi Sawatani, Director of the IT Division’s Consumer Support System Development Department in the Business Support System Division at SoftBank Corp., says, “The way they work is very analog, so their workload onsite has become quite heavy.”

“In 2015, there were very few smartphone advisers in the roughly 2,500 SoftBank shops nationwide,” Sawatani continues. “By 2018, there were still only about 600. But in 2019, with the sense that 600 was not enough to serve growing demand, SoftBank increased that number to 1,200 advisers. In parallel to this rapid payroll growth, people working in the shops began to urge that their manual work be systematized.”

Staff in SoftBank. shops interacted with customers via a business app installed on iPad devices. However, the smartphone advisers had no business app. Instead, they had to use multiple paper-based systems to make reservations for smartphone classes, manage customer status, and handle their other responsibilities, resulting in poor efficiency.

To meet these demands, the IT Division embarked on a path to create a faster solution. An agile development system was put in place, welcoming members on the business side—who directly interacted with smartphone advisers—into the development team.

“The company has predominantly used waterfall-style development, which starts coding only after requirements are specifically defined,” says Sawatani. “SoftBank had never used anything like the new approach, in which team members from the business side described the type of system they wanted. To conduct all the interviews needed to finalize requirements would probably have meant six months or more for development. To avoid spending that kind of time, it was important for us to adopt a process of ‘build it and try it,’ working with business-side team members to move forward quickly.”?

The platform they chose for this sort of agile development was?Microsoft Azure Kubernetes Service (AKS), a fully managed Kubernetes service provided on the?Microsoft Azure?public cloud platform hosted by Microsoft.

Shigeru Tatsuta, Senior Engineer, Consumer Support System Development Department, Business Support System Division, IT Division, SoftBank Corp.

Tsuyoshi Sawatani, Director, Consumer Support System Development Department, Business Support System Division, IT Division, SoftBank Corp.

Yoshio Terada, Java Evangelist, Microsoft Cloud Developer Advocate, Microsoft Japan

SoftBank maximizes Azure advantages by making full use of PaaS functionality

In fact, SoftBank had already adopted cloud services from Microsoft, and it used?Azure DevOps?as a project management tool for agile development even before the IT Division started talking about developing the Smartphone Adviser Reservation Tool (SMART) application. According to Shigeru Tatsuta, Senior Engineer in the Consumer Support System Development Department at SoftBank Corp., that gave the team great familiarity with AKS as a managed Kubernetes service.

“The first thing we thought of doing was implementing a system that lowers the load on the operator by using Kubernetes to automate system operation work,” says Tatsuta. “We had multiple options within the company for the infrastructure on which we could develop the solution—most importantly, our proprietary data center. However, with no Kubernetes specialist onsite, it would require a lot of work for us to build our own on-premises environment, including learning to install it and preparing a server. And should there be any problem in the Kubernetes environment, the company had no site reliability engineer (SRE) in-house, meaning we had to figure out who would restore the environment. In that respect, Azure provided all the services we needed to implement DevOps, including Azure DevOps and Azure Kubernetes Service (AKS). Using AKS was the fastest shortcut for a developer-led tryout of Kubernetes.”

In July 2019, the company held a four-day “hack fest” with the theme of using AKS and Azure DevOps to mount a single function within a system that SoftBank shop staff would use from iPad devices. The hackfest confirmed the effectiveness of doing so.

A hackfest is a method of selecting a single problem and working to solve it together, in this case, with top-class engineers from Microsoft Japan. Yoshio Terada, a Java Evangelist and Microsoft Cloud Developer Advocate, hosted the hackfest. He recounts, “The people from SoftBank’s IT Division were passionate and quick to understand. Even before the hackfest began, they’d resolved more than I had imagined they could. It went faster than any hackfest I’ve ever led.”

Tatsuta describes what the hackfest taught him about the capabilities of Azure.

“When our IT Division first embarked on agile development, we used an agile project management tool called Mingle from ThoughtWorks, but its service life was drawing to a close,” recalls Tatsuta. “When we looked for a replacement, we found Azure DevOps and immediately tried it, discovering that its project management tool was very well done. It also provided a CI/CD experience via Azure Repos and Azure Pipelines. And by combining it with Azure Monitor and Application Insights, we found we could make system monitoring easy—for example, by getting logs that span multiple clusters. Naturally, there was no talk of?not?deploying these specifically designed functions, but since they could not be used in an on-premises environment, we needed to use Azure. We took on the challenge of building our app as a cloud-native service using Azure. We ran SMART in parallel with our development work for about one year, including the release of the trial version. We found that the system’s simple operation tasks could be greatly reduced by automating them using Kubernetes. Taken to the extreme, this meant developers could even monitor operations while developing the system. We felt impelled to implement DevOps. I realized that these were the advantages of using a cloud-native platform as a service (PaaS).”

Improvements on two-week cycles meant development wrapped up quickly

With this background work done, developing the SMART business app for smartphone advisers went smoothly. Product development happened in a series of two-week sprints. In December 2019, on the eighth sprint, SoftBank released a trial version to select users. Apart from skipping the sprint at the start of April 2020 due to the COVID-19 emergency declaration, the IT Division updated the app’s functionality in each of these sprints. SoftBank had released SMART in an ongoing series of stages starting December 2019, and by the time the emergency declaration was lifted in June 2020, service had been extended to 1,200 smartphone advisers throughout Japan.

During this development period, “It was very interesting to get feedback that could only be obtained by actually building the software and then having stakeholders use it,” recounts Tatsuta. “We struggled most with improving usability. All that struggle paid off, though, in bigger results. When we had the smartphone advisers actually use what we developed, they generated new ideas on the spot. Getting feedback about things that the engineering side had not even considered led to radical improvements. What made me happiest was converting the paper consent forms that customers used to sign into digital versions. Signing paper forms had created the task of storing documents securely in a vault, and paper storage gives rise to its own security concerns. By providing functionality for digital forms, we helped resolve those issues. Smartphone advisers in the stores were delighted. Some aspects of implementing the electronic signature part took us into the technological unknown, but it was worth it.”

Now, customer information can be located and instantly called up using contract telephone numbers as the key. When the relevant smartphone adviser is not present, staff members in the shop can still smoothly work with the customer.

“We are still adding features and improvements,” says Tatsuta. “Even after the system went into service in June, the sprints continued. Development is going strong on new features that will be released soon.”

Using Kubernetes to automate operational tasks keeps the discussion lively

Tatsuta describes how this project renewed his sense of the effectiveness of Kubernetes. “The environment we built strikes a balance with security policies,” he says. “At the moment, we cannot implement the linked use of AKS virtual nodes and Azure Pipelines—which use Azure Container Instances—with AKS as we originally envisioned. But even so, by building a cloud-native system using Azure, we made its operation much more efficient. For SoftBank, we like Kubernetes more as a platform for automating operational work than as a platform for implementing microservices. In fact, based on the results achieved in the SMART project, the discussion with our IT Operating Division is deepening as we explore the way system operations will change in the future. This includes DevOps. The experience has been a real opportunity for transformation. I think Azure Kubernetes Service (AKS) has been a really good way to get our feet wet with Kubernetes.”

Sawatani adds, “This project was a challenge for us. However, thanks to our experience with agile development using new technology, we were able to present a case study to conferences outside the company and achieve a good result in relation to our target, which was to train engineers in advanced skills used outside SoftBank. Our goal for SMART development is in sight, and a turning point awaits. While we have not decided what role Kubernetes will play for us in the future, we envision a series of transformations in our company in the post-COVID-19 age. The new knowledge we have accumulated as engineers will certainly be valuable. We hope that Microsoft continues to add and improve features.”

---------------------------------------------------------------------------------------------------------------Completed Task-21 in the program ARTH-The School of Technologies.

要查看或添加评论,请登录

Sridhar Rao Kukkala的更多文章

  • KUBERNETES: CASE-STUDY

    KUBERNETES: CASE-STUDY

    What is Kubernetes? Kubernetes is a portable, extensible, open-source platform for managing containerized workloads and…

    1 条评论
  • AMAZON SQS: USE-CASES

    AMAZON SQS: USE-CASES

    What is Amazon Simple Queue Service? Amazon Simple Queue Service (Amazon SQS) offers a secure, durable, and available…

    3 条评论
  • MongoDB: CASE-STUDY

    MongoDB: CASE-STUDY

    What is MongoDB? MongoDB is a document-oriented NoSQL database used for high-volume data storage. Instead of using…

    6 条评论
  • OPENSHIFT : CASE STUDY

    OPENSHIFT : CASE STUDY

    What is Red Hat OpenShift? OpenShift is a family of containerization software products developed by Red Hat. Its…

  • Jenkins : Case Study

    Jenkins : Case Study

    What is Jenkins? Jenkins? is an open-source automation server. With Jenkins, organizations can accelerate the software…

  • Role of Neural Networks in Tesla

    Role of Neural Networks in Tesla

    What are Neural Networks? Neural networks are a set of algorithms, modeled loosely after the human brain, that is…

    1 条评论
  • AI_ON_AWS SelfReflection

    AI_ON_AWS SelfReflection

    In workshop regarding AI_ON_AWS conducted by Vimal sir discussed the following topics Amazon Comprehend : Advanced text…

  • How Microsoft solving challenges using Ansible....

    How Microsoft solving challenges using Ansible....

    What is Ansible Ansible is a software tool that provides simple but powerful automation for cross-platform computer…

  • Self-Reflection of Practical Demo session by Experts from RedHat

    Self-Reflection of Practical Demo session by Experts from RedHat

    In the program of ARTH-The School Of Technologies attended Expert's session Practical Demo of Enterprise Level…

    1 条评论
  • How Apple using AI / ML

    How Apple using AI / ML

    What is Artificial Intelligence? Artificial Intelligence is the process of creating a human-like intelligence that can…

社区洞察

其他会员也浏览了