Completed 25PB large data migration in 2 years using 252 AWS Snowball Edges

2022

KDDI CORPORATION. provides network solutions as a comprehensive telecommunications carrier. The company migrated its back-end system from an on-premises environment to Amazon Web Services (AWS) for its mobile communication service au users. AWS Snowball Edge with physical appliances were utilized to complete the migration of 25PB in about 2 years. The migration along with the activation of the storage tier optimization function led to 50% savings in operating costs.

AWS Case Study | KDDI CORPORATION
kr_quotemark
The deciding factor in adopting AWS was the technical value and cost benefits that were seen during the planning phase of the AWS Migration Acceleration Program (MAP) of migration assistance program. The PoC with AWS Snowball Edge convinced us that we managed to achieve the migration of 25PB of data which we initially thought was impossible.

Kensuke Tsukishima
KDDI CORPORATION 
Vice President, Engineering Division
Vice President, Technology Strategy & Planning Division

Migration to AWS for flexible resource expansion of storage services

KDDI focuses on mobile and Internet services in Japan under the philosophy of "contributing to the development of an affluent communications society.” In addition to the relationship with AWS as a user of cloud services, we also collaborate with AWS on business aspects such as providing AWS Wavelength which enables the construction of ultra-low latency applications within au's 5G network.

KDDI's migration to AWS this time is for the platform of its "Data Storage Service" provided to au users. Data storage service is a storage service for photos, videos, address books, calendars, etc. taken with smartphones. au Smart Pass / au Smart Pass Premium members can use up to 50GB and non-members can use up to 1GB.

The data storage service which has operated in an on-premises environment since launching year of 2012 was required to optimize server and storage resources as the number of users increased. Mr. Takamitsu Katayama of the Engineering Division, Application Engineering Department said,
“The amount of data in photos and videos has increased dramatically due to the increasing sophistication of cameras. In the on-premises environment, servers and storage have been expanded about twice a year, but it was difficult to calculate the required capacity and the operational load was increasing. Therefore we have decided to migrate to cloud services to enable flexible resource expansion, optimize storage costs, and improve customer experience through enhanced performance."

For cloud services, AWS was selected based on a comprehensive evaluation of serverless and container services that provide high resource efficiency, managed services that enable fundamental cost improvements and the ability to balance performance and cost through automatic scaling.

Mr. Kensuke Tsukishima, Vice President, Engineering Division and Vice President, Technology Strategy & Planning Division said, "The deciding factor in adopting AWS was that we managed to see the technical value and cost benefits in the planning phase through the AWS Migration Acceleration Program (MAP) of migration support program. The PoC with AWS Snowball Edge convinced us that we managed to achieve the migration of 25PB of data, which we initially thought was impossible."

Innovation cycles reduced to 40 days by using 252 AWS Snowball Edge devices

The project to migrate to AWS began in October 2018. The trials for the 25PB data migration using AWS Snowball Edge took place from May to July of 2019, with the data migration piece from August to July of 2021.

AWS Snowball Edge is a service that stores data in a dedicated physical appliance and transports it to AWS. In this case, the company used 18 AWS Snowball Edge units per cycle for the migration which was repeated for 14 cycles (252 total).

For the data migration, we prepared 18 units per cycle including two spare units for two sets of 18 units for a total of 38 units and operated the two cycles in parallel. To speed up the data transfer from the on-premises environment to AWS, Snowball Edge in-memory processing was also implemented and helped improve throughput.

“As a result, we were able to reduce the data transferring time to AWS Snowball Edge to be around 40 days per cycle and migrate 25 PB of data in about two years. During the migration, we also worked closely with the Technical Account Manager (TAM) on AWS' enterprise support team and received feedback as appropriate on potential enclosure failures, migration progress, and how to proceed. Thanks to this, we were able to proceed with the project without anxiety and shorten the construction period" said Katayama.

Modernize architecture by leveraging serverless and containers

The new platform for data storage services was released in two phases in April 2019 and August 2020. In Phase 1, where the system was migrated from on-premises, AWS Lambda enabled event-driven thumbnail generation in a serverless manner in addition to system migration by lift. In Phase 2, we shifted to a cloud-native architecture by moving the DB from Amazon RDS to Amazon Aurora and the container execution environment from Amazon EC2 to AWS Fargate.

At the end of Phase 2, we have worked to optimize costs and introduced Amazon S3 Intelligent-Tiering, a tiered storage service in March 2021.

"With features such as the ability to automatically move data that has not been used for a long period of time to a more cost-effective tier, we were able to curb the ever-increasing costs of Amazon S3 and migrate to AWS without sacrificing the customer experience. We were able to reduce our storage operating costs by about 50%. As new storage tiers are being updated and usage fees are being lowered, we would like to take advantage of this to improve our cost efficiency. “ said Katayama.

Releasing new services for users through effective use of engineering resources

The migration to AWS has ensured scalability and flexibility to handle an ever-increasing number of users and data. Storage costs which account for the majority of system operations, it has been reduced by approximately 50% and contributing to a reduction in TCO. We no longer have to scale up or down servers or spend lead time adding storage capacity which is our biggest concern," said Tsukishima.

The serverless and managed services have also greatly reduced the operational burden. Mr. Nobutaka Suzuki as Engineering Division, Department manager, Application Engineering Department said, "As a result, we were able to shift engineering resources to the development of new services which led to the release of the “Data Moving Service.” Engineers are more motivated and there is more momentum to actively take on the challenge of DevOps style development"

We plan to continue our efforts to modernize and make our systems cloud-native in order to optimize costs and enhance operational sophistication as well as to maintain and improve service quality. “We see potential in a multi-region architecture that leverages the Osaka region of AWS to improve service quality. We expect AWS to continue to provide new services and help us reduce costs," said Suzuki.

The Engineering Division of the Technology Management Head Office who is responsible for the maintenance and operation of the entire telecommunications service stated that the know-how gained through from this migration will be also horizontally deployed to support mission-critical services.

“We aim to provide high quality services by realizing a hybrid environment, utilizing on-premises and cloud computing in the right places at the right time. We would like to build a good relationship with AWS while maintaining a cooperative structure," said Tsukishima.

Kensuke Tsukishima
Vice President, Engineering Division
Vice President, Technology Strategy & Planning Division

Nobutaka Suzuki
Engineering Division, Department manager, Application Engineering Department

Takamitsu Katayama
Engineering Division, Application Engineering Department


Customer Profile: KDDI CORPORATION

  • Date of Established: June 1, 1984
  • Capital: JPY 141,852 million
  • Number of employees: 47,320 (as of March 31, 2021)
  • Business activities: Telecommunications business

Benefits of adopting AWS and future prospects

  • Migration of 25PB of large data in about 2 years
  • Reduces storage operating costs by about 50%
  • Reduced lead time for procurement and expansion of resources
  • Promote modernization and cloud-native systems

Key Services Currently In Use

AWS Snowball Edge

Snowball Edge is an edge computing and data transfer device powered by the AWS Snowball service. It has onboard storage and computing capacity and offers select AWS services for use in edge locations.

Learn more »

AWS Lambda

AWS Lambda lets you run code without provisioning or managing servers. You pay only for the compute time you consume.

Learn more »

Amazon ECS

Amazon Elastic Container Service (Amazon ECS) is a fully managed container orchestration service. Customers such as Duolingo, Samsung, GE, and Cook Pad use ECS to run their most sensitive and mission-critical applications to gain security, reliability, and scalability.

Learn more »

AWS Fargate

AWS Fargate is a serverless computing engine for containers that runs on both Amazon Elastic Container Service (ECS) and Amazon Elastic Kubernetes Service (EKS). With Fargate, you can easily focus on building your application.

Learn more »