Swarm is designed to provide both fast and secure data storage. Selecting the appropriate trade-offs between performance, capacity, and data protection that best meet your business needs is a critical part of a successful storage solution.
Swarm includes the Health Processor (HP) that provides end-to-end, disk-level, and life-cycle data protection. The Health Processor monitors both data integrity and cardinality (the number of object replicas) continuously, and it heals any degradation or non-conformity within the storage cluster that it finds.
The Health Processor regularly scans every object on disk to verify its integrity. If the object is corrupt (such as due to a bad sector), HP removes it. HP then detects and corrects the object's under-replication by triggering the creation of another replica of that object elsewhere in the cluster, thus restoring the correct number of replicas.
Using the Health Processor, Swarm can:
Verify the correct object protection level exists in the cluster.
Distribute replicas and erasure coding segments properly across subclusters.
Enforce lifepoint policies by enforcing replica and/or erasure coding segment counts and delete policies (terminal lifepoints) at different policy time intervals.
Validate the object on-disk integrity and create a new replica if the integrity is compromised.
Economically store and load balance by periodically evaluating if an object is optimally stored in its current location.
Defragment storage space on an as-needed basis.
See Lifepoint Metadata Headers.
In addition to the protection provided by the Health Processor, Swarm uses the Content-MD5 metadata header to provide end-to-end message integrity check of the object body (excluding metadata) as it is sent to and from Swarm.