About services
A service is a background process that performs a specific function that contributes to the continuous tuning of the HCP system. Services work across all namespaces in the repository.
In general, services run only while they are enabled. An exception is the Protection service, which runs in response to certain triggers even while it is disabled. Typically, services are disabled only by authorized HCP service providers during problem resolution.
The Overview page of the System Management Console shows the status of most services. HCP records information about service runs and irreparable violations in the system log.
Service types
A number of services contribute to the continuous tuning of the HCP system.
HCP implements the following services:
Capacity Balancing service
Ensures that the percentage of space used remains roughly equivalent across the storage nodes in the HCP system.
Compression/Encryption service
Compresses object data to make more efficient use of HCP storage.
Content Verification service
Ensures that object data is not corrupted.
Disposition service
Automatically deletes expired objects.
Duplicate Elimination service
Merges duplicate data to free space in the HCP storage. You can monitor the activity of this service.
Fast Object Recovery service
Ensures that unavailable objects have their status changed to available once they are recovered.
Garbage Collection service
Deletes data and metadata left in the repository by incomplete operations, thereby freeing space for the storage of additional objects.
Geo-distributed Erasure Coding service
Erasure codes full copies of object data in replicated namespaces that allow erasure coding.
Migration service
Migrates data off selected nodes in an HCP RAIN system or selected storage arrays in an HCP SAIN system in preparation for retiring those devices.
Network Per Storage Component service
Increases tiering performance from the HCP system to HCP S Series Nodes or external storage devices by isolating their communication to an individual forward-facing HCP network. Each HCP S Series Node or external device can use its own network to communicate with HCP.
Protection service
Ensures that damaged or lost objects can be recovered.
Replication service
Maintains selected tenants and namespaces on two or more HCP systems and manages the objects in the selected namespaces across those systems to ensure data availability and enable disaster recovery. You can configure, monitor, and control the activity of this service.
Replication Verification service
Replicates objects that the Replication service missed replicating or was unable to replicate.
HCP S Series Balancing service
Ensures that the percent of space used across S Series Nodes in a pool remains roughly equivalent.
Scavenging service
Ensures that the metadata for each object exists and is not corrupted.
Shredding service
Shreds deleted objects that are marked for shredding.
Storage Tiering service
Moves objects among storage tiers, creates and deletes copies of objects on various storage tiers to ensure that each tier contains the correct number of copies of each object, and changes objects to metadata-only according to rules in service plans.
Service precedence
Some services take precedence over others.
- On any given node, the Capacity Balancing service takes precedence over the Scavenging service. If the Scavenging service is running when the Capacity Balancing service starts, the Scavenging service stops. When the Capacity Balancing service stops, the Scavenging service automatically restarts, providing that it is scheduled to run at that time.
- On any given node, the Migration service takes precedence over the Capacity Balancing service. If the Capacity Balancing service is running when the Migration service starts, the Capacity Balancing service stops. It does not restart automatically when the Migration service stops.
- On any given node, the Protection service takes precedence over the Compression/Encryption and Content Verification services. If either of these services is running when the Protection service starts, the service that was running stops. When the Protection service stops, each service that stopped automatically restarts, provided that the service is scheduled to run at that time.
Metadata storage
To fully understand how certain services work, you need to know how HCP manages metadata.
When you add an object, upload part of a multipart object, or add a directory or symbolic link to a namespace, the following events occur.
- HCP creates primary metadata for the item being added. This metadata consists of information HCP already knows, such as the creation date, and for objects and parts only, the data size, hash algorithm, and cryptographic hash value generated by that algorithm. It also includes metadata that was either inherited or specified in the write request, such as retention setting, UID, and GID.
- HCP creates a second copy of the primary metadata. HCP then distributes both copies of the primary metadata among the HCP general nodes.
- For objects and parts of objects:
- HCP creates the number of copies of the object data or part data required to satisfy the ingest tier data protection level (DPL) in the service plan associated with the namespace. If the ingest tier is primary running storage, HCP distributes all copies of the data among the HCP storage nodes. If the ingest tier is HCP S Series storage, HCP writes all copies of the data to that storage.
Each copy of the primary metadata for the object or part points to all copies of the data for that object or part. However, object or part data in primary running storage is not necessarily stored on the same nodes as the primary metadata for the object or part.
- In primary running storage, HCP stores the number of copies of the object or part metadata required to satisfy the ingest tier metadata protection level (MPL) in the service plan. These copies, called secondary metadata, let HCP reconstruct the primary metadata should that become necessary.
If the ingest tier is primary running storage, the MPL is the same as the DPL.
- HCP stores MPL copies of any custom metadata for the object in primary running storage. For multipart objects, HCP stores MPL copies of the custom metadata for the object as a whole instead of MPL copies for each part.
- If the ingest tier is HCP S Series storage, HCP stores one copy of the secondary metadata along with each copy of the object data on that tier.
- HCP creates the number of copies of the object data or part data required to satisfy the ingest tier data protection level (DPL) in the service plan associated with the namespace. If the ingest tier is primary running storage, HCP distributes all copies of the data among the HCP storage nodes. If the ingest tier is HCP S Series storage, HCP writes all copies of the data to that storage.
The following figure shows the data and primary and secondary metadata that result from storing an object in a namespace with a service plan that sets both the DPL and MPL for the ingest tier to 2. The figure assumes that the ingest tier is primary running storage.
When an object or part is moved from the ingest tier to another storage tier, HCP stores one copy of the secondary metadata along with each copy of the object or part data on that tier. Regardless of which tier the object or part data is on, HCP always keeps in primary running storage the number of copies of the secondary metadata and custom metadata required to satisfy the MPL for that tier.