Dell PowerScale Deploy flashcards

72 cards. Read the term and the definition here. The deck at the top of this page is optional.

OneFS namespace

A single distributed file system namespace spanning all nodes; every node can serve any file, and clients see one consistent directory tree regardless of which node holds the data.

Protection level +2d:2n

Reed-Solomon erasure coding with two data and two parity blocks per stripe; tolerates two simultaneous node or drive failures and requires at least three nodes in the cluster.

Usable capacity formula

Raw capacity multiplied by N divided by N plus M, minus metadata, snapshot reserve, and workflow reserve; for example 10 nodes of 10 TB with +2d:2n yields about 80 TB before reserves.

SmartConnect zone

A group of external interfaces and floating IPs with a delegated DNS name; the cluster answers A records with available node IPs so clients load balance and survive node failure.

Backend network isolation

The internal network between nodes (InfiniBand fabrics or dedicated Ethernet switches) must be non-routable and separate from client and management traffic, with consistent MTU across all nodes.

DNS delegation for SmartConnect

An NS record delegates a subdomain to the cluster; without it, SmartConnect zone names do not resolve and clients cannot use zone-based load balancing or failover.

Access zone

A partition of the namespace with its own IP pool, authentication providers, SMB shares, and NFS exports; the System zone is reserved for administration and cannot be deleted.

File pool policy

A rule that matches files by path, name, type, size, or timestamps and assigns protection level, storage pool tier, and snapshot settings; policies are evaluated in order.

Rolling OneFS upgrade

An upgrade performed node by node while the cluster stays online; all nodes must end on the same version, and release notes list supported paths and firmware prerequisites.

SMB continuous availability

A share setting combined with SMB3 that lets clients reconnect transparently after a node failover; only CA-aware clients benefit, and the share must be marked CA.

NFSv4 idmap

The service that maps Windows SIDs to POSIX UIDs and GIDs; misconfiguration causes files to appear owned by nobody or access to be denied for NFSv4 clients.

OneFS S3 service

An object API where buckets map to directories, access keys authenticate users, and bucket policies control access; it is not full AWS feature parity, so some features are unsupported.

Snapshot reserve

Space set aside for copy-on-write snapshots; when snapshots exceed the reserve they spill into free space and can fill the file system, so usage must be monitored.

SyncIQ failover

A process that makes the target cluster writable after replication; failback uses a change list to resync divergent data before normal replication resumes.

isi_gather_info

A support bundle command that collects logs, configuration, and diagnostics from the cluster; generate it before changing configuration when escalating an issue.

SMB share permissions

Share-level and file-level ACLs combine, and the effective permission is the intersection; both must grant access for a user to read or write the file.

NFS security flavors

Export options sys, krb5, krb5i, and krb5p; krb5p adds privacy through encryption at a performance cost, and Kerberos requires correct time sync and keytabs.

Hard versus soft quota

A hard quota blocks writes at the limit, while a soft quota only sends notifications; default quotas apply to new directories created under a parent directory.

SmartConnect load policies

Policies such as round-robin, CPU usage, network throughput, and connection count determine which node IP a client receives; choose based on whether the workload uses many small or few large connections.

IP pool rebalancing

Floating IPs move between nodes when a node fails or load is uneven; a rebalance policy and interval control how aggressively the cluster redistributes addresses.

Replication bandwidth throttling

SyncIQ traffic can be placed on a dedicated subnet and limited by bandwidth rules so replication does not starve client workloads during large transfers.

isi status

A command that reports cluster health, node states, and restriping progress; a node in down or smartfailed state triggers rebuild and reduces redundancy until complete.

isi statistics

Commands that report protocol, drive, node, and client statistics; high latency with low throughput often points to network or client-side problems rather than storage.

NDMP backup

A backup protocol requiring a DMA server and a tape or disk target; verify the NDMP service is enabled and credentials match, and correlate DMA and cluster logs on failure.

Node generation families

F-series is all-flash for performance, H-series is hybrid for mixed workloads, and A-series is archive for cold data; generation suffixes indicate newer hardware, not protocol changes.

OneFS REST API

The platform API exposes configuration, statistics, and provisioning over HTTPS; automation tools and Dell integrations use it for repeatable cluster management.

isi healthcheck

A suite of checks on configuration, network, and data protection; run it after installation and after major changes to catch misconfigurations early.

AD join prerequisites

DNS must resolve the domain, NTP must keep clocks within tolerance, and a service account with join rights is required; an optional OU can control where the computer object is created.

S3 bucket policy

A policy attached to a bucket that grants or denies access to users and accounts; the S3 service must be enabled per access zone, and access keys authenticate requests.

Kerberos clock skew

Kerberos rejects tickets when client and server clocks differ by more than about five minutes; verify NTP on clients, nodes, and the domain controller before debugging keytabs.

Redundancy states

Healthy means full protection, degraded means a drive or node is missing and rebuild is underway, and critical means another failure would cause data loss.

SyncIQ policy monitoring

Commands list policies and reports showing last sync status, lag, and errors; investigate a failed policy before attempting failover to avoid data loss.

CloudPools

A licensed feature that tiers cold data to cloud object storage such as AWS S3 or Azure Blob while keeping the namespace local; recalls happen on access.

Protection level +3d:3n

Erasure coding with three data and three parity blocks per stripe; tolerates three simultaneous failures and requires at least four nodes, reducing usable capacity more than +2d:2n.

IP pool sizing

Provide at least one floating IP per node per subnet so failover has an address to move; more IPs improve load distribution across nodes.

LACP aggregation

Link aggregation configured per interface group on the external network; the switch port-channel must match the node's aggregation mode to avoid asymmetric hashing and drops.

Cluster configuration wizard

The first-boot workflow that collects cluster name, external and internal network settings, DNS, NTP, and admin credentials; other nodes join and receive the configuration over the internal network.

Storage pool tiers

Pools group drives by type, such as SSD, HDD, or archive; file pool policies place data on a tier based on matching criteria and set protection for those files.

NFSv4.1 sessions

NFSv4.1 adds sessions and pNFS support; OneFS supports it for modern clients, while NFSv3 remains common for legacy workloads that do not need the newer features.

Access zone isolation

Each zone has its own authentication and IP pool, so a client in one zone cannot see another zone's shares unless explicitly configured; this supports multitenancy.

Drive smartfail

OneFS marks a failing drive as smartfailed and rebuilds its data using parity; replace the drive and verify with isi status that rebuild completes and redundancy returns to healthy.

isi event list

A command that shows recent events by severity; correlating drive, network, and protocol events helps distinguish a single failure from a cascading issue.

Snapshot restore

A single file or directory can be restored from a snapshot without reverting the whole file system, using the .snapshot directory or the UI; test restores before relying on snapshots.

SmartQuotas

A licensed feature enforcing directory, user, and default quotas; hard limits block writes, soft limits notify, and reports show usage per quota domain.

Node count and protection

+2d:2n needs at least three nodes, +3d:3n at least four, and +4d:4n at least five; choose protection after planning node count, because it determines usable capacity and failure tolerance.

Management network separation

The management network carries the OneFS UI and API, while IPMI is a separate out-of-band path; keeping them off the data subnet supports security and isolation policies.

License activation

Feature licenses such as SnapshotIQ, SyncIQ, SmartQuotas, CloudPools, and S3 are activated through the UI or API before configuring dependent policies or services.

Policy evaluation order

File pool policies are evaluated in order and the first match applies; place specific policies before general ones so files land in the intended pool with the intended protection.

SMB3 encryption

An SMB3 feature that encrypts data in transit; it protects sensitive traffic but can reduce throughput, so test performance impact before enabling cluster-wide.

HDFS NameNode proxy

A proxy address that clients use for high availability instead of individual node IPs; it fails over between nodes and shares the namespace with NFS and SMB access.

isi space reports

Reports showing used, available, and snapshot space; unexpected growth often comes from snapshots or SyncIQ staging, so check both before adding capacity.

Interface error checks

Review interface status, errors, drops, and LACP state on the node side, and confirm DNS resolution of SmartConnect zones before blaming storage for slow client performance.

Cluster as management unit

OneFS is managed at the cluster level, not per node; configuration, policies, and monitoring apply across all nodes, and the cluster is the unit of upgrade and licensing.

File striping with parity

Files are striped across nodes with parity blocks, so a file's data lives on multiple nodes; node loss does not lose data when the protection level tolerates that failure count.

Aggregate client bandwidth

The sum of client throughput should not exceed external network capacity; use multiple high-speed links and LACP, and plan for headroom during failover.

SyncIQ link sizing

Replication bandwidth and recovery point objective determine link sizing; throttle replication traffic so it does not compete with client workloads during large transfers.

Hardware compatibility check

Before installation, verify the node model and drive firmware are supported by the target OneFS release; unsupported combinations can block node join or upgrade.

Test restore and replication

After configuring snapshots and SyncIQ, perform a test restore and a test replication before relying on them; untested protection is not proven protection.

S3 access keys

Per-user credentials for the OneFS S3 service; bucket policies can grant cross-account access, and the S3 service must be enabled in the relevant access zone.

Permission inspection

Use share and export listing commands plus idmap checks to inspect effective permissions; share and file permissions intersect, and UID/GID mismatches cause denials.

Node failure triage

Check IPMI, power, and internal network links for a down node; a down node triggers rebuild if protection allows, and removal must follow the proper procedure.

SyncIQ change list

A record of changes used during resync after failover; it reconciles divergent data between source and target before normal replication resumes.

Node generations and tiers

F-series is all-flash for IOPS-sensitive workloads, H-series is hybrid for mixed use, and A-series is archive for cold data; mixing types in a cluster affects policy placement.

Protection level +4d:4n

Erasure coding with four data and four parity blocks per stripe; tolerates four simultaneous failures, requires at least five nodes, and has the lowest usable capacity ratio.

SmartConnect zone design

Use one zone per subnet or tenant, delegate the zone name in DNS, and avoid overlapping existing records; zones should not span unrelated security domains.

Backend MTU consistency

All nodes and switches on the Ethernet backend must use the same MTU, typically 9000; inconsistent MTU causes fragmentation, drops, and slow inter-node traffic.

Access zone IP pool

Each zone is assigned an IP pool from a subnet; the pool supplies floating IPs for SmartConnect load balancing and failover within that zone.

Snapshot schedule

A SnapshotIQ configuration setting directory, interval, retention, and reserve usage; test a restore from the schedule before depending on it for recovery.

SMB1 default state

SMB1 is disabled by default in current OneFS releases; enabling it for legacy clients weakens security and should be limited to the specific hosts that require it.

Quota notification

Thresholds on a quota trigger alerts or scripts when usage crosses a percentage; notifications support capacity planning before a hard limit blocks writes.

isi healthcheck

A command that runs configuration, network, and data protection checks; results highlight misconfigurations that may not appear in normal status output.

Support escalation bundle

Generate isi_gather_info, record the event timeline, note the OneFS version, and list recent changes before contacting support to speed diagnosis.