Přeskočit na obsah

Our development

Beside operating the infrastructure and providing it to researches across different scientific areas, there is also a need to maintain and further develop its "background" services. These services help infrastructure administrators to cope with various infrastructure-related issues (e.g., scheduling subsystem, security and monitoring, clouds and virtualizations, etc.). Of main interest are also services that serve infrastructure users (e.g., PBSMon, Perun, etc.).

Software developed in MetaCentrum can be divided into following categories.

Clouds and Scheduling

  • Torque Batch System - job management system. Fork of the original Torque batch system that was greatly enhanced for stability and to cover all features provided by the Czech NGI. The system is coupled with a custom high-performance queue based scheduler.
  • HPC Cloud - MetaCloud is a computing (HPC) cloud built using Amazon compatible cloud interface.
  • Alea Simulator - advanced, open source, platform independent job scheduling simulator designed for study, testing and evaluation of various job scheduling techniques. This simulator is capable to realistically simulate complicated job scheduling problems using real or artificial workload and failure traces. Alea provides both textual and graphical simulation outputs and includes implementations of several common scheduling algorithms. New algorithms can be added through provided interfaces.
  • OCCI Standard Development and Support - participation in OGF's OCCI Working Group to develop the Open Cloud Computing Interface; support OCCI for server- and client-side applications through the rOCCI framework and jOCCI Java libraries.

Identity Managment and Security

  • Perun - CESNET cooperates on design and development of Perun system used for managing users, groups and sccess to resources in distributed environments. Perun has been integrated with a number of infrastructures and projects on national and international level.
  • Handling and Prevention of Security Incidents - in order to face new threats, MetaCentrum develops tools and procedures to efficiently handle security incidents and to detect them quickly. We are also part of EGI security teams that are responsible for the security operations of the EGI infrastructure.
  • Beyond Identity Federations - based on experience with a world-wide used set of clinical and histological medical images, we have developed a concept of user-centric identity federations. We piloted non-web federation technology Moonshot with several use-cases.

Complex Monitoring and Accounting

  • PBSmon - a web-based framework for monitoring and accounting in the Czech NGI. It visualizes the current state of computing resources, Torque (PBS) state and user presonalized job information.
  • Pakiti - service Pakiti is developed to detect machines that are not properly updated with security patches. Utilizing the Pakiti service site administrators can detect unpatched machines, which otherwise pose a significant risk for the whole infrastructure.

 

 

 

Last changed:Mon Dec 12 14:21:41 CET 2016