MySQL案例研究-KDDI使用MySQL InnoDB Cluster防止服务宕机
免费下载
1 CASE STUDY | MySQL Enterprise Edition | KDDI Corporation
Copyright © 2020, Oracle and/or its affiliates
KDDI prevents service downtime with MySQL
InnoDB Cluster and reduces failure recovery time
by 80%
“MySQL InnoDB Cluster automates failover on database
failure. Anyone can perform the same high-quality operation
independent of the skill level of the operations staff.
No service down due to the database has ever occurred.”
KDDI Corporation
Platform Development Division
Service Application Development Department
Framework Group Supervisor
Yusuke Suzuki
CASE STUDY
KDDI Corporation is increasing in-house DBAs (Database Administrators) and aims
to internalize implementation and operation of the database to avoid system failure.
The company has adopted MySQL InnoDB Cluster and MySQL Enterprise Edition to
strengthen database quality control.
INTRODUCTION
KDDI is a comprehensive telecommunications carrier that provides services such as
“au” in mobile communications and fixed/international communications. KDDI has
set “being a company that customers can feel closest to” as one of its goals, and is
expanding business in various areas other than telecommunications. With the slogan
of “the integration of telecommunications and life design”, KDDI is spreading
services from payments to e-commerce and social infrastructure, including Cashless
payment “au PAY,” internet banking “Jibun Bank,” IoT service “au HOME” that makes
life easier and more convenient by operating home appliances with smartphone
apps, and electricity business “au Denki.”
KDDI’s Service Application Development Department, Platform Development
Division, builds the IT infrastructure that supports communication services and non-
communication businesses..
The Service Application Development Department has various organizational
functions, and KDDI DBA Secretariat is positioned as a CoE (Center of Excellence)
specialized for databases.
KDDI DBA Secretariat is responsible for consistent quality control of about 1,000
system databases in the company, working closely with many development and
operation departments. Specifically, the Group provides training and technical
support for the DBAs who are in charge of the development and operation of each
service, standarding parameters, monitoring items, operations, and providing tools
for development.
CHALLENGE
• In 2014-2015, KDDI assigned DBAs to each development and operation division
and they are supported by members dedicated to the Framework Group
databases. The reason KDDI tried this system was that there were many failures
KDDI CORPORATION
Platform Development Division
Service Application Development
Department
Framework Group Manager
Shinya Miyazawa
KDDI CORPORATION
Platform Development Division
Service Application Development
Department
Framework Group Supervisor
Yusuke Suzuki
2 CASE STUDY | MySQL Enterprise Edition | KDDI Corporation
Copyright © 2020, Oracle and/or its affiliates
caused by databases at that time. Analysis of those troubles revealed the
following two major problems.:
• First, the number of people who were familiar with the databases in the company
was limited. When a failure occurred, they spent too much time resolving the
trouble because they didn't know the details of the database, which was like a
“Black box.”
• Second, KDDI didn't have enough resources to check every item entrusted to SI
partners and keep the quality of database to a high level because KDDI had many
SI partners whose skills with databases were different, and this caused the first
problem as well.
• KDDI DBA Secretariat created a basic policy on quality control for the databases of
about 1,000 systems that exist in the company, and believed that two database
management and operational policies were needed.
• One is a system managed individually and totally like the “F1 machine.” The
database supporting various telecommunication services, which is KDDI's core
business, requires a platform with higher performance and availability, such as
Oracle Exadata Database.
• The other is a system generalized and integrated, such as “a mass‐produced car.”
It was difficult to estimate the future workload of databases generated by services
in System of Engagement (SoE) fields other than telecommunications, so KDDI
often launched the systems on a small start basis. KDDI uses MySQL provided by
Oracle for these platforms.
• However, KDDI needed to guarantee the availability and reliability even for
general-purpose and collectively managed databases. The cost was an essential
requirement at the same time as guaranteeing the service level of the databases.
KDDI believed MySQL InnoDB Cluster would be a powerful solution for these
types of database. KDDI also decided to adopt MySQL Enterprise Monitor and
MySQL Enterprise Backup from MySQL Enterprise Edition.
RESULTS
• By the end of June 2019, KDDI had launched two systems using MySQL InnoDB
Cluster.
• One is a system originally using MySQL Master-Slave configuration for
replication. The rapid increase in service demanded higher system requirements
than initially expected. Since migrating to another platform while providing
services was too risky, KDDI chose MySQL InnoDB Cluster as the best option to
ensure availability under the same MySQL architecture.
• The other is a system construction project which was started to shift to in-house
development. KDDI aimed to accelerate service deployment by using MySQL
InnoDB Cluster.
• When a failure occurred in the conventional MySQL Master-Slave configuration,
the operator had to perform many operations for recovery, but the number of
operators having the necessary skills was limited. MySQL InnoDB Cluster solved
the problem and realizes automatic failover, eliminating the fear of a long service
outage.
• MySQL Shell has been particularly useful in disaster recovery. It enables everyone
to perform the same high-quality operations without depending on the skill level
of the operator and reduces the time required for recovery by more than 80%. In
some cases, by using MySQL Shell, failure recovery time has been reduced to
about 10 minutes when it used to be nearly an hour.
• MySQL InnoDB Cluster made it possible for KDDI to recover from potential failure
without affecting applications and businesses. No database downtime has occurred
since the system was launched.
• In addition, MySQL Enterprise Monitor is very effective in monitoring failures and
detecting potential issues. For operators who are not database experts, it was
of 4
免费下载
【版权声明】本文为墨天轮用户原创内容，转载时必须标注文档的来源（墨天轮），文档链接，文档作者等基本信息，否则作者和墨天轮有权追究责任。如果您发现墨天轮中有涉嫌抄袭或者侵权的内容，欢迎发送邮件至：contact@modb.pro进行举报，并提供相关证据，一经查实，墨天轮将立刻删除相关内容。
下载排行榜
评论