Clusters and Pinning Snapshots

Cluster creation

This release supports cluster creation: a "cluster" is a special definition incorporating a head node and a worker node context.

Clusters have the following features:

• they automatically scale based on their load;
• they can be deployed very easily by copying and pasting an auto-generated one-liner to the command prompt.

It is possible to encrypt clusters for being able to store sensitive data securely.

Pinning Snapshots

It is possible to select which revision of the CernVM Filesystem to use when launching a CernVM virtual machine. The list of available snapshots is directly taken from the repositories.

The advantages are twofold:

• Consistency: it is possible to run an entire cluster of virtual machines with the same version of CernVM, instead of always picking the latest each time a new VM is started.
• Long-Term Data Preservation: it will be possible in the future to restore the exact environment where a certain data processing occurred today, by greatly reducing the hassles of OS/software incompatibility when reprocessing "ancient" data.