RUBYKAIGI2014

Detail

[JA] Resource Control Architecture scripting with mruby for a Web Server Separating Computer Resources Virtually at Each HTTP Request

As the increase of large-scale Web services, Web applications are increasingly sophisticated. The increase of smartphones makes traffic to Web servers increase each day. Management of Web servers is increasingly important. When a Web server receives resource-hungry requests, administrators of the web server must control the resources flexibly. However, existing resource control methods process mandatorily such as denial, disconnect or the like of the request when the number of simultaneous connections, request per unit time, CPU used time or memory utilization exceed a certain threshold value. These methods can’t control resources processing request. In this situation, if one request has the potential to occupy a majority of computer resources, the server process can process other request in principle by separating resources of the server process temporarily without using up the saving of computer resources. In this paper, we propose a resources control architecture scripting with mruby for a Web server separating computer resources virtually by a HTTP request using mod_mruby and mruby-cgroup. Each HTTP request processing is allocated virtual computer resources temporarily and the request is process within the limit of the allocated computer resources in our architecture. Therefore, each request processing has a very small effect if clients access to contents heavily or a particular content occupy a majority of computer resources. If a request has the potential to occupy a majority of computer resources, our architecture can process the request continually within the limit of the allocated computer resources without denying or disconnecting requests mandatorily.

Recorded Video

Resource Control Architecture scripting with mruby for a Web Server Separating Computer Resources Virtually at Each HTTP Request from rubykaigi on YouTube .

Presentation Material(s)

MATSUMOTO Ryosuke

Company
Kyoto University
Job title
Ph.D. Student
Bio

After getting bachelor, he worked for three years in a hosting company. Experience in the company was accepted, and he skip the master's program and admmition to doctoral program in the Graduate School of Informatics, Kyoto University. He has been researched about OS, Middleware and protocol. He interested in security, resource management, performance, large-scale, operation technology for Web servers, the next generation infrastructure technology like Test-Driven Infrastructure and Immutable Infrastructure, and the next generation HTTP protocol like SPDY, HTTP/2 and QUIC in particular. He is mod_mruby, ngx_mruby and Trusterd HTTP/2 Web Server author. 2013, IPSJ IOTS2013 Best Paper Award, 2014, 6th FUKUOKA Ruby Award Outstanding Performance Award, 9th Japan OSS Encouragement Award.

Location
Osaka, Japan

Back to SCHEDULE