|HOME page for
The portal is dedicated to keep the references concerning sites
where information about Linux clusters is.
Now most of activity on the site is moved to
- A computer cluster is
a collection of the interconnected computers (computing machines) under
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:
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
- 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
data transfer tools between clusters.
- To keep cluster up we need to be informed about all types of security:
All above questions are represented on this Security: all aspects ....
- 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)
- 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
- 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 ....
- 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.