Netapp NS0-077 - NetApp StorageGRID Administrator Exam

The correct architecture combines S3 versioning, cross-grid replication (CGR), and ILM retention. StorageGRID cross-grid replication operates between two StorageGRID systems connected through grid federation and requires the participating source and destination buckets to have versioning enabled. Critically, CGR creates versioned objects at the destination and replicates the source version ID, preserving the ordering of object versions between the grids. This directly satisfies the requirement to protect all possible versions.

After replication, each grid independently applies its active ILM policy to the objects stored locally. The backup grid can therefore use ILM rules to retain replicated objects for at least 90 days. StorageGRID ILM also supports rules specifically targeting noncurrent versions of versioned S3 objects by using Noncurrent time as the reference time, allowing previous versions to be retained for the required period.

Options A and B are unsuitable because nonversioned buckets cannot preserve every object version. Option C correctly identifies versioning and CGR, but a bucket policy governs access permissions; version retention is instead controlled through ILM or an S3 lifecycle configuration.

Reference topics: Grid Federation → Cross-Grid Replication → S3 Versioning → ILM Rules → Noncurrent Object Versions → Retention Periods

The correct answer is Gateway Node. StorageGRID 12.0 provides load balancer caching, which is specifically designed to improve performance when workloads repeatedly access the same subset of objects. NetApp identifies AI/ML training as a strong candidate for this feature because training jobs often reread the same datasets across multiple epochs. By serving frequently requested objects from cache, StorageGRID can reduce backend object retrieval activity and lower response latency.

NetApp explicitly states that load balancer caching is available only on Gateway Nodes. Administrators enable caching on a load balancer endpoint associated with Gateway Nodes and should use dedicated cache-enabled endpoints for workloads that benefit from repeated reads. NetApp also lists data lakes, high-performance computing, AI/ML training, content distribution, media asset management, and video production as examples of workloads that can benefit from caching.

Option B is incorrect because Storage Nodes hold object data and metadata but do not provide the integrated load-balancer caching capability. Option C is incorrect even though Admin Nodes run the Load Balancer service; the caching feature itself is restricted to Gateway Nodes. Archive Nodes are unrelated to this functionality.

Therefore, D. Gateway is the verified answer.

Reference topics: StorageGRID 12.0 → Gateway Nodes → Load Balancer Caching → Cache-Enabled Endpoints → AI/ML Training Workloads

A healthy StorageGRID deployment requires one primary Admin Node and at least three Storage Nodes at each site. NetApp StorageGRID 12.0 documentation states that every grid requires a primary Admin Node to provide core management and configuration services, while every site requires a minimum of three Storage Nodes. Therefore, the smallest healthy single-site virtual test environment contains four nodes: one Admin Node plus three Storage Nodes.

The three-Storage-Node minimum is important because Storage Nodes host core distributed services such as Cassandra, DDS, and ADC. In particular, the ADC service is hosted on at least three Storage Nodes at a site, supporting grid topology information, authentication, and service discovery.

Option A lacks the required primary Admin Node and therefore cannot constitute a complete manageable StorageGRID grid. Option D contains an Admin Node but only one Storage Node, which does not satisfy the three-node minimum. Option C would work, but the Gateway Node is optional because Admin Nodes can also provide S3 load-balancing functionality for a small compatibility-test deployment.

Therefore, B is the minimum healthy virtual configuration.

Reference topics: StorageGRID Architecture → Grid Node Types → Primary Admin Node → Minimum Storage Nodes per Site → Virtual Deployment Requirements

The correct configuration is Cross-Origin Resource Sharing (CORS). StorageGRID supports CORS specifically so that objects stored in an S3 bucket can be accessed by web applications originating from different domains. NetApp's documentation uses essentially this exact use case: an S3 bucket containing images can be configured with CORS so that those images can be displayed by a website hosted in another domain.

For a service provider hosting image objects centrally while multiple customer websites access those objects from separate domains, the browser treats the requests as cross-origin. A CORS configuration on the S3 bucket defines the permitted origins, HTTP methods such as GET, and allowed headers. The administrator can either authorize specific customer domains or, where appropriate, permit GET requests from multiple origins. HTTPS remains handled independently by the StorageGRID load balancer endpoint and its TLS certificate.

Option C is incorrect because the HTTPS certificate must match the StorageGRID endpoint domain names clients connect to, not every external website that merely references the images. Option A is unnecessary because one endpoint can service multiple domains, and Swift is unrelated.

Reference topics: Tenant Manager → S3 Buckets → Cross-Origin Resource Sharing (CORS) → Allowed Origins → Static Web Content → HTTPS S3 Access

The required consistency level is strong-global, specifically using the Quorum strong-global semantics available in StorageGRID 12.0. NetApp defines strong-global as guaranteeing read-after-write consistency for client requests across all sites. With Quorum semantics enabled, a grid containing three or more sites can tolerate a site failure while continuing client operations, provided the required metadata replica quorum remains available.

This directly matches the scenario. NetApp's multisite resiliency guidance states that, for a three-or-more-site deployment using Quorum strong-global, a single site plus a single node failure can occur without interruption to operations or metadata consistency, assuming sufficient object-data protection remains under the ILM configuration.

Option A, all, is unsuitable because every metadata node must acknowledge the update; if a node or site is unavailable, the request fails. Option C, strong-site, only guarantees immediate consistency within one site and does not provide the required global consistency. Option D provides read-after-write consistency for new objects but only eventual consistency for updates, so it does not satisfy the requirement that data remain consistently available for all read and write activity.

Reference topics: StorageGRID 12.0 → S3 Consistency → Strong-global → Quorum Semantics → Multisite Failure Tolerance → Metadata Replication

The correct answer is metadata updates. StorageGRID maintains object metadata independently from object data, using the distributed Cassandra database. NetApp specifies that StorageGRID stores three copies of the metadata for every object at each site, meaning every site maintains metadata for all objects in the grid, regardless of where the object data itself is physically stored.

This distinction is critical in the scenario. Even though the new fourth site uses an ILM policy that keeps its object data local with a 2+1 protection scheme, metadata activity is still grid-wide. NetApp explicitly states that administrators must consider network bandwidth and latency between sites because metadata updates are continually replicated between sites even when all object copies remain at the site where they were ingested. Therefore, the WAN links connecting all four sites must be sized to support continuous metadata synchronization.

Load balancer endpoints handle S3 client connectivity and do not create this universal intersite requirement. NTP requires reliable time synchronization but is not the workload driving the stated bandwidth-sizing requirement. Storage pools are logical ILM constructs and do not themselves replicate metadata.

Therefore, B. metadata updates is the required component.

Reference topics: StorageGRID Expansion → Adding a Site → Cassandra Metadata Replication → Intersite Network Bandwidth and Latency → ILM Data Placement

StorageGRID load balancer endpoints can be restricted so that only selected tenant accounts are permitted to use a specific endpoint. During endpoint configuration, the administrator uses the Tenant access setting and selects Allow selected tenants, then chooses the backup tenant account. NetApp states that with this setting, only the selected tenant accounts can use the endpoint to access their buckets. This directly satisfies the requirement to dedicate one S3 load balancer endpoint to the internal-backup tenant.

Option A is incorrect because a global binding mode determines where the endpoint is available across eligible Admin and Gateway Node interfaces; it does not restrict access by tenant identity. Option D similarly limits the endpoint to interfaces associated with particular HA groups, which controls network exposure rather than tenant authorization. Option B is not a valid per-tenant isolation mechanism because HTTPS configuration applies to the endpoint connection itself and cannot be selectively disabled for unrelated tenant accounts.

Therefore, the required control is to configure the load balancer endpoint's tenant-access settings and assign access specifically to the backup tenant.

Reference topics: Grid Manager → Load Balancer Endpoints → Tenant Access → Allow Selected Tenants → S3 Client Access Control → Endpoint Binding

For a capacity-focused FabricPool repository containing predominantly cold object data, the most efficient expansion method is to add a data-only Storage Node. StorageGRID 12.0 supports combined, metadata-only, and data-only Storage Nodes. A data-only node is dedicated to object-data storage and does not participate in Cassandra metadata storage, making it particularly appropriate when the primary objective is to maximize usable object capacity rather than increase metadata capacity. NetApp specifically describes data-only, high-capacity Storage Nodes as appropriate when dedicated capacity is required.

After the node is added, erasure-coded data can be rebalanced so that existing EC fragments are redistributed across the original and newly added Storage Nodes. NetApp states that EC rebalancing redistributes erasure-coded fragments among Storage Nodes at the site, allowing newly added capacity to be utilized by an existing EC protection scheme.

Option D adds usable capacity, but a combined node also provides metadata/Cassandra functionality that is unnecessary when the goal is specifically to maximize object-data capacity for a cold FabricPool workload. Options B and C are not supported StorageGRID capacity-expansion strategies; appliance RAID/DDP configuration and SANtricity LUN sizing must not be arbitrarily changed to enlarge StorageGRID object storage.

Reference topics: StorageGRID Expansion → Data-only Storage Nodes → Object Capacity → Erasure Coding → EC Rebalancing → FabricPool Capacity Planning