Using MHA (Master High Availability Manager for MySQL) and Murakumo (distributed internal dns) to achieve a simple highly-available environment.
#autotechcasual
株式会社ミライタスの社内発表会でG(インフラグループ:サーバー)チームが発表した「Linux Server 冗長化~リアルタイム同期でラクラク運用~」ののスライドです(個人名を含む部分等、一部変更を加えています)。専門的な内容ですが、実際にメンバーの1人が作業を行い検証したため、手順書としても使用できるものとなっています。
株式会社ミライタスの社内発表会でG(インフラグループ:サーバー)チームが発表した「Linux Server 冗長化~リアルタイム同期でラクラク運用~」ののスライドです(個人名を含む部分等、一部変更を加えています)。専門的な内容ですが、実際にメンバーの1人が作業を行い検証したため、手順書としても使用できるものとなっています。
Infinite Debian - Platform for mass-producing system every secondTaisuke Yamada
Starting from standard install and various "Debian internal" for mass-installation system, the talk goes on to describe generic "instant system generation" which shortens turnaround time from 10s of minutes to seconds. All based on Debian.
Prepared for kickstart meetup of FukuokaDebian.
Infinite Debian - Platform for mass-producing system every secondTaisuke Yamada
Starting from standard install and various "Debian internal" for mass-installation system, the talk goes on to describe generic "instant system generation" which shortens turnaround time from 10s of minutes to seconds. All based on Debian.
Prepared for kickstart meetup of FukuokaDebian.
This document summarizes a talk given by Michael H. Oshita on December 8, 2014 about AWS Lambda. The talk discussed Oshita's background and experiences using AWS Lambda, including issues invoking Lambda functions from SNS and reviewing logs to debug problems. The conclusion warns about needing to be careful with escapes when using AWS Lambda.
This document discusses the setup of a Varnish server for a Ruby on Rails application. It includes components like Ruby on Rails, Redis, Varnish with ESI and gzip, and Nginx. It also discusses using Edge Side Includes (ESI) to reduce requests to the backend, A/B testing functionality using Varnish Control Language (VCL), and the Lacquer gem for Rails integration with Varnish.
The document discusses benchmarking and tuning a web application architecture to improve performance. It analyzes GET and POST request performance, tunes Apache MaxClients and InnoDB settings, and benchmarks the effects of adding APC, PHP optimizations, and caching with Varnish. It then benchmarks load balancing with HAProxy and LVS, and integrating Varnish, Lua, and Memcached caching to further optimize response times.
This document discusses setting up a custom non-RDS MySQL replication configuration across multiple Availability Zones on AWS. It describes using MHA (Master HA Manager) to monitor a MySQL master and promote a slave to master if the original master fails, allowing for high availability and automated failover. Non-RDS configurations offer more flexibility compared to RDS by allowing custom storage engines, distributions, instance types and fast (<20s) failover without DNS issues.
1) The document discusses the creation of a website called sinsai.info to aggregate earthquake information from various sources using open source tools like Ushahidi during the 2011 Tohoku earthquake and tsunami in Japan.
2) Performance issues arose as traffic increased, including high latency and database overload. Solutions involved optimizing the database, adding an Elastic Load Balancer, and backing up to Amazon S3 and EBS.
3) The website served as a useful resource for earthquake information, and the author thanks AWS for its services which helped handle the increased traffic and issues during the disaster.
This document discusses using HAProxy to provide high availability for MySQL databases running on Amazon EC2. It describes setting up a MySQL master-master replication configuration across two EC2 instances with HAProxy load balancing between the databases. HAProxy is configured to monitor the MySQL servers and direct reads to an available master while allowing writes to both masters for redundancy.