Cohesity DataProtect flashcards
Cohesity DataProtect · Build recall with a focused review session.
32 cards. Read the term and the definition here. The deck at the top of this page is optional.
- SpanFS
- Cohesity's distributed filesystem that spans all nodes into a single namespace, storing immutable fingerprinted chunks and re-protecting data after node or disk failure.
- View
- A logical data container on a Cohesity cluster exported over NFS, SMB, or S3; snapshots and protection groups operate on views and their contents.
- Helios
- Cohesity's SaaS management plane for multi-cluster visibility, policy, and reporting; the data path stays on-cluster, so backups continue during Helios outages.
- Protection Source
- A registered workload endpoint such as vCenter, a physical server with an agent, a NAS share, a database, or a cloud account, requiring least-privilege credentials to enumerate and snapshot.
- Protection Policy
- Defines schedule, retention, and replication or archival targets; a protection group binds sources to a policy to determine when and how data is copied.
- CBT reset
- A changed-block-tracking reset forces the next backup to read all data as a new full; expect it after VM migration or disk changes.
- Application-consistent backup
- A backup that quiesces the application, using VSS on Windows or pre/post scripts on Linux, so the captured data is transactionally clean and recoverable.
- SLA policy
- An objective-driven policy such as bronze, silver, or gold that derives run frequency, retention, and copies automatically to meet the stated service level.
- Instant recovery
- Runs a workload directly from the backup store for the fastest RTO, then migrates it to production storage to complete the recovery.
- Synthetic full
- Assembles a new full backup from prior incrementals without re-reading the source, reducing source load and storage churn during incremental-forever backups.
- Log backup
- Captures database transaction logs for SQL or Oracle, enabling point-in-time recovery to a chosen second; a broken log chain limits recovery to the last good log.
- Instant volume mount
- Presents a backup as a datastore or LUN to an ESXi host, allowing fast recovery of a VM whose data runs from the cluster until migrated.
- Job warning
- A job state where some objects were skipped; those objects remain unprotected until the next successful run, so warnings are actionable, not success.
- Immutability
- A retention lock or WORM policy that prevents deletion or modification of snapshots until expiry, even by an administrator, defeating ransomware that targets backups.
- Compliance mode
- A retention-lock mode where the retention period cannot be shortened by anyone; governance mode instead allows a privileged override.
- Air-gap replication
- Replicating backups to a remote cluster in a separate security domain so a compromised primary cannot delete the secondary copy.
- Anomaly detection
- Monitors backup metadata and behavior, flagging sudden data reduction or unusual deletion and encryption patterns that may indicate ransomware before retention is consumed.
- Audit log
- Records who deleted or changed snapshots, policies, and users; retained as compliance evidence and used in incident investigation.
- Rebuild headroom
- Reserved spare capacity, commonly around 20 to 30 percent, that lets a cluster re-protect data after a node failure without degrading production performance.
- iris_cli
- The on-cluster command-line interface used for status, service restart, statistics, and node-level diagnostics not always exposed in the graphical interface.
- NTP skew
- Clock drift between nodes or services that breaks Kerberos authentication and replication; correct time synchronization before deeper troubleshooting.
- Support bundle
- A collected set of node and service logs gathered before opening a support case; correlate timestamps across nodes to find the root cause.
- RPO
- Recovery point objective, the maximum acceptable data loss; set by backup frequency, log backup interval, and replication lag.
- RTO
- Recovery time objective, the maximum acceptable downtime; set by the recovery method, such as instant recovery versus full restore.
- Replication factor
- The number of copies stored for each chunk; usable capacity is roughly raw capacity divided by the factor, before erasure coding and reserve.
- Erasure coding
- A data protection scheme that spreads parity across nodes, reducing capacity overhead compared with full replication while tolerating node or disk loss.
- SmartTiering
- Moves cold snapshots to cheaper archive media while metadata stays on-cluster, freeing capacity and enabling recall when the data is needed.
- Witness node
- A lightweight node used in stretched or remote-office topologies to maintain cluster quorum without storing full production data.
- Tag-based protection
- Dynamic source selection where new virtual machines matching a defined tag are automatically included in a protection group without manual registration.
- Point-in-time recovery
- Restores a database to a specific second by replaying transaction logs from a base backup; requires an intact and unbroken log chain.
- Orphaned snapshot
- A hypervisor snapshot left behind when removal fails during backup; it grows the snapshot chain and must be consolidated before the next run.
- Separation of duties
- A control where backup administrators do not hold domain administrator rights, limiting the blast radius if backup credentials are stolen.