Terminology and components
- Bin
- A container for data of zero or more slice names that occupies sequential space on disk and typically could be written and read in a single IO operation. Slices are assigned to bins based on a slice name hash value. Note Potentially, multiple slices could share a bin, which is known as a hash collision. The number of bins on disk is large enough to avoid frequent hash collisions.
- Catalog (one instance per disk)
- A database that holds extant slice names and their revisions. This database allows for efficient client object slice listing operations.
- In-memory pointers
- All bin locations are stored in memory; these in-memory pointers are used when servicing normal client IO. The amount of pointers that can be held depends on available system memory at the time of storage pool creation.
- Journal
- An internal data structure to maintain ZSS specific information utilized to recreate ZSS state on startup. While the ZSS is running all changes to disk metadata are appended to the journal.
- Zone
- An addressable portion of a drive that allows only sequential writes on a 4K boundary. A zone size is defined for the entire drive but could be different for different drives
- Zone Slice Storage
- Zone Slice Storage (ZSS) is a new form of long-term storage that adopts many principles from PSS but makes design improvements in several key areas. ZSS removed the externally provisioned file system that managed data placement and replaced it with an internal data structure. This helped to reduce overhead improving general write and read performance. All writes utilize an append-only construct similar to PSS, but instead of bin files created by a filesystem, ZSS writes to a zoned format on the hard disk. Similar to PSS the idea of compacting areas to clean up any holes left by deleted data is still there, but this is done more efficiently using new algorithms because ZSS controls data placement. ZSS provides stronger than PSS consistency and recovery guarantees in case of crashes, power outages and malicious actions and byte level utilization accuracy Like Be the first to like this
- ZSS instance
- A set of zones on a single drive. One is deployed separately per drive. There are as many instances as there are active drives in a running Slicestor node.