clustergate

HOME page for ClusterGateOrg

The portal is dedicated to keep the references concerning sites where information (in English and Russian) about computing clusters is.

A computer cluster is a collection of the interconnected computers (computing machines) under united administration. It is used when one machine is not enough to accomplish calculations in reasonable time or (more general) when one machine is not enough to serve requests from users. It may be considered computing cluster, database cluster or http cluster, etc.
Here we plan to bring together various useful information about computer clusters. It is planned to represent the information mainly about midrange clusters with use of open source software. The authors are first users of this portal.
- Several wide areas were and are important for us:
  - General Linux: general information about operating system Linux itself and examples of Linux distributions.
  - The information how to do clustering. Which software is used to keep a computer cluster up, to configure it, etc.
  - Also important general purpose system software which could be used on a cluster. Where are the resources for such free software.
  - Hardware is real base for computer cluster. We mean computers as well as communication components (may be Ethernet, Myricom, InfiniBand, etc.), disk drives, and other cluster related hardware components.
  - When we discuss speed of I/O we need to have a lot of info about file systems of various types: local, distributed, virtual, etc. In most cases we need to make the speed for data reading/writing as much as possible.
  - Having discussed the cluster topics we have to mention access methods to the large volume of data and tools for data transfer between clusters mainly over WAN.
    - To keep cluster up we need to be informed about all types of security:
      - network security;
      - data integrity and safety, i.e. to have backup copy for serious cases (for example disk drive is broken, etc.);
      - of course we have to remember Uninterpretable Power Supply units and appropriate control when electricity power is going down (how to do proper system shutdown in time)
    - All above questions are represented on this chapter: Security: all aspects ....
  - How to know what is going on on the cluster exactly, on every machine and every Ethernet port, and many other details? Of course it could be done with special software packages, so called monitoring and measurement tools.
    - How to balance the load between the machines in the cluster? Obvious desire is make the machines in the cluster loaded evenly (it gives minimum turn around time for job performance). Several of popular balance systems are mentioned on the page Batch/Load Balance systems
  - Another interesting point is virtualization: how to port almost any software service from one machine to another one in minutes and how to use one machine for many virtual machines.
  - May be most interesting topic is real life examples for a range of clusters: High Performance Clusters, midrange clusters, other examples are explored on the pages: High Performance Computing: Examples of powerful clusters, Examples: midrange clusters, and also Computing in High Energy Physics: Computing sites, application packages.
  - Finally we could not forget cluster of clusters, so called Grid .... Also we keep in mind virtualization and clouds.
  - Information about computer clusters are growing every day. Clusters are used more and more often. The page Further reading: Journals, Reviews, News, Books gives more information on the issue.

Andrey Ye. Shevel