Tokyo-based telecom company SoftBank, has tapped software- defined storage to support business-critical IT while deliver
ADOPTING A REVOLUTIONARY STORAGE STRATEGY Tokyo-based telecom company SoftBank, has tapped softwaredefined storage to support business-critical IT while delivering cost savings, flexibility and the scalability to accommodate future growth
Telecommunications
|
Japan
Business needs
Solutions at a glance
Facing enormous pressure to reduce costs in the face of slow-to-no-growth mobile markets, SoftBank wanted to continue to simplify its operations and standardize on hardware while deploying flexible, scalable storage that reduces its expenses, ensures high performance and availability, and allows for future expansion.
• Dell EMC ScaleIO software • VMware suite of products • Linux KVM • OpenStack
Business results • Delivers CapEx and OpEx savings to help SoftBank cut costs in half
• Offers performance that meets throughput and latency targets
30
%
lower CapEx costs
• Ensures high availability to support missioncritical applications
• Scales to cost-effectively support future growth of IaaS/cloud and container workloads
4.1 PB across three data centers
A subsidiary of Japanese technology giant SoftBank Group Corp., SoftBank Corp. is a leading multinational telecommunications company and service provider. SoftBank owns and operates a wide range of businesses — including mobile communications, fixed-line telecommunications and broadband internet. The company is facing numerous challenges. Mobile subscription growth has slowed dramatically in Japan as the market reaches saturation. To address this, SoftBank has implemented a program it calls “Half and Twice” — with the objective of cutting half of its costs while doubling productivity. For the IT infrastructure team, this has meant finding innovative ways to reduce its operations expenses while delivering the exceptional performance required to satisfy the service quality expectations of its customers worldwide. “Slashing our operations costs by half wouldn’t be possible without revolutionary changes to our entire approach to infrastructure,” says Yasuyuki Kato, senior director of Systems Infrastructure Division of SoftBank’s IT division. “Dell EMC ScaleIO was a key element in bringing this about.”
Capitalizing on commodity hardware Over the years, SoftBank has purchased tens of thousands of servers and more than 300 arrays as a storage area network (SAN) to support its expanding operations. To manage its capital investment, the company chose to standardize on lower-cost, x86 servers that it could buy in large quantities from various vendors to obtain the best pricing. This also simplified the IT team’s design efforts and reduced its maintenance and support costs. To leverage its investment in this commodity hardware while obtaining the storage performance needed for mission-critical IT operations, SoftBank decided to implement software-defined storage. Over a two-year period, the company conducted an extensive review of providers. It chose ScaleIO not only for the functionality and simplicity it provided, but also for its extraordinary
flexibility and scalability, and the support that Dell EMC offers. “With the hardware layer standardized and unified on commodity servers, software became the critical storage component,” Yasuyuki explains. “With ScaleIO, we were able to effectively create storage area network appliances on the servers without needing to buy any hardware or even dedicated storage enclosures.”
“Slashing our operations costs by half wouldn’t be possible without revolutionary changes. Dell EMC ScaleIO was a key element in bringing this about.” Yasuyuki Kato Senior Director, Systems Infrastructure Dept. IT Division, SoftBank
An innovative clustering approach Scales to costeffectively support
future
growth of IaaS/cloud and container workloads
SoftBank used ScaleIO to consolidate block storage for its internal IT as part of its new infrastructure-as-a-service (IaaS) and private cloud platform. Before ScaleIO, storage was deployed using legacy network storage appliances, each supporting as many as 100 virtual machines (VMs). Although managing the VMs with traditional SAN appliances seemed easy, it quickly became apparent that when anything went wrong with the underlying storage layer, all the related VMs went down together. SoftBank hedged against this risk by storing volume data on the local disks of each node. But it then faced the possibility of losing data availability should a host server node go down. SoftBank’s solution was to build out a large number of small, six-node ScaleIO clusters in a hybrid configuration. All six nodes contribute to the ScaleIO storage pooling, while five of the six also host application VMs — the sixth one functioning as a “hot spare” for the VMs should one of the first five fail. Further, all the VMs within an application cluster were then placed onto separate ScaleIO-backed clusters, ensuring continued application availability even if an entire ScaleIO cluster went down.
“With ScaleIO, we were able to effectively create storage area network appliances on the servers without needing to buy any hardware.” Yasuyuki Kato Senior Director, Systems Infrastructure Dept. IT Division, SoftBank
“By doing this, we were able to take advantage of virtualization and abstraction for both our compute and storage clusters,” notes Yasuyuki. “At the same time, we ensured high performance and continuous operation for business-critical applications such as our billing system, using multiple ScaleIO and VM host-node clusters.” SoftBank has also deployed ScaleIO storage-only clusters, providing block storage to both VMware and OpenStack compute clusters. Because OpenStack is a crucial part of the company’s infrastructure, ScaleIO’s integration with OpenStack met a key requirement in its adoption decision.
Cutting costs while maintaining high performance Deploying over 4 PB of ScaleIO across three data centers, SoftBank reduced its CapEx by 30 percent for implementation costs alone. What’s more, because ScaleIO essentially provides all-flash storage while allowing the company to continue to buy commodity servers at volume discounts — with a return-for-replacement policy rather than having to spend for expensive, on-site spares — the overall savings have helped SoftBank meet its goal of cutting operations costs in half.
“Because ScaleIO requires minimal host CPU resources, we believe that in the near future ScaleIO on ARM servers will be a great match.” Yasuyuki Kato Senior Director, Systems Infrastructure Dept. IT Division, SoftBank
ScaleIO’s ability to auto-balance data across the clusters and to self-heal should a node go down further enhances SoftBank’s TCO. In addition, ScaleIO’s REST APIs enable unprecedented automation through simple integration with SoftBank’s orchestration solution.
ARMing itself for further cost savings Going forward, SoftBank continues to look for ways to better manage costs while continuing to deliver the quality of service that will enable it to profitably expand its customer base. In 2016, the SoftBank Group acquired ARM Holdings — a large, UK-based designer and licenser of RISC processor chips used in mobile phones, tablets and the coming generation of smart IoT products. SoftBank believes that lower-cost ARM host servers can be used for ScaleIO.
Learn more about Dell EMC solution
“Because ScaleIO requires minimal host CPU resources, we believe that in the near future ScaleIO on ARM servers will be a great match,” Yasuyuki remarks. “We could then move our ongoing business applications to ARM server platforms as appropriate. We are working closely with a number of people including Dell EMC developers to create storage that combines ARM and ScaleIO.”
Contact a Dell EMC Expert
Copyright © 2017 Dell Inc. or its subsidiaries. All Rights Reserved. Dell, EMC and other trademarks are trademarks of Dell Inc. or its subsidiaries. Other trademarks may be trademarks of their respective owners. This case study is for informational purposes only. The contents and positions of staff mentioned in this case study were accurate at the point of publication, November 2017. Dell and EMC make no warranties — express or implied — in this case study. Reference Number: H16630.
Connect on social
