Exam Information
Free interactive sample
1Z0-997-26 Short Quiz
10 Real Exam Questions - One at a Time - Instant or End-of-quiz Scoring.
An e-commerce company hosts a stateless application on an instance pool across three availability domains in a single region. Autoscaling is configured on CPU utilisation. During a flash sale, the pool scaled to its maximum size, and the load balancer began returning 503 responses. The architect finds that one availability domain had no capacity for the requested shape at the time and the pool could not place instances there, so the remaining domains were overloaded. The company wants the application to remain available during zonal capacity shortages without raising the minimum footprint dramatically. Which change is the most appropriate?
Regional subnets let instances be placed across availability domains, and offering alternative shapes gives the pool flexibility when one shape is short on capacity in a given domain. Raising the maximum or lowering the threshold does not fix placement failure when a domain cannot host the shape at all. Changing the load balancer type does not address the instance supply problem, and a second pool in the same domain shares the same capacity constraint.
A platform team runs several microservices on OKE and wants a central, cost-conscious observability design. They need to route all error-level application logs from the cluster to a long-term Object Storage archive, and they want to find, across all compartments, the compute instances tagged for a specific project without knowing their compartment. Which TWO actions together best meet these requirements? (Choose two.)
A Connector Hub connector with a log source filter routes only the required error-level records to a durable Object Storage target, which keeps archive costs under control. OCI Search lets a query span resources across compartments by type and tag, which answers the discovery need. Logging Analytics dashboards are for visualisation rather than archival, a Streaming export is not a long-term archive, and Monitoring metric queries do not return resource discovery results by tag.
An operations team needs to be alerted when the 5xx error rate of a load balancer exceeds 2 percent over five minutes, and the alert must go to the on-call Slack-connected topic. A dashboard showed a spike, but no notification arrived. The team confirmed that the load balancer emits a metric for HTTP responses in the Monitoring namespace. Which configuration step is most likely missing?
An alarm can be fully configured and still notify no one if the topic it targets has no active subscription for the on-call endpoint, which matches the symptom of a spike with no notification. The raw-count concern is a query design issue rather than the most likely cause given the stated metric. Audit logging of alarm state is not required to deliver notifications, and five-minute windows are a valid evaluation setting.
A global SaaS provider needs its customer data to be served with low read latency in three continents. Writes must be accepted in a single home region for regulatory reasons, and reads from other regions must reflect committed data within seconds. The architect considers Globally Distributed Autonomous Database. Which design element is most important to get right to meet the write-location and read-latency requirements?
Globally Distributed Autonomous Database keeps a primary region for write traffic while serving reads from replicas and shards nearer to users, which is why the connection-string routing matters to meeting both requirements. Independent per-continent databases with application replication would break the single-home-region write requirement and create consistency problems. A CDN cannot serve transactional database reads, and adding OCPUs in one region does not reduce cross-continent latency.
A development team is containerizing an Oracle Functions workload and an OKE-hosted service that share a build pipeline. They want the pipeline to build, push, and deploy both artefacts with the same credentials, and they want no long-lived credentials stored in the pipeline. The security team also requires that the deployment stage can only update resources in the production compartment. Which TWO actions together best satisfy these requirements? (Choose two.)
A dynamic group with a compartment-scoped policy lets the pipeline authenticate without stored credentials while keeping its permissions confined to production. Using resource-principal connections from the DevOps pipeline to OCIR and OKE provides the single identity for both artefacts without long-lived secrets. A tenancy-wide admin user with a stored key violates both the no-long-lived-credentials and least-privilege requirements, root-level grants exceed the production-only restriction, and committing a token to the repository exposes a secret.
A government agency runs workloads in a compartment hierarchy managed by OCI Organizations. The security team must guarantee that no one in any child tenancy can disable Cloud Guard or create public Object Storage buckets, even tenancy administrators of those child tenancies. A central team wants to apply this consistently across twelve accounts. Which mechanism best enforces this requirement?
Service control policies in OCI Organizations set guardrails that apply to every principal in the attached tenancies, so they cannot be overridden by local administrators. Local IAM policies can be edited by tenancy administrators, which defeats the requirement. Read-only access does not block the change, and responder rules react after the fact rather than preventing the action.
A team uses OCI Resource Manager to deploy a stack that provisions a VCN, an OKE cluster, and an Oracle Functions application. They want a change to the stack to be previewed before anything is modified, and they want the deployment to be reproducible across the development and production tenancies from the same Terraform configuration. During a plan run, the output shows that the OKE node pool would be replaced because a shape attribute changed. The production team wants to avoid unplanned node replacement during business hours. Which action is the best next step?
Resource Manager separates plan and apply jobs, so the replacement can be reviewed in the plan output and the apply can be scheduled into a maintenance window. Parameterising environment-specific values as stack variables keeps one configuration reproducible across tenancies. Auto-approving during business hours causes exactly the disruption the team wants to avoid, editing the state file by hand is unsafe and bypasses the managed state, and creating resources manually defeats the reproducibility requirement.
A healthcare provider runs a three-tier application in the Ashburn region: a web tier behind a load balancer, an application tier on compute instances, and an Oracle Base Database Service system. The recovery objective is a Recovery Time Objective of 30 minutes for the entire stack, and the team must rehearse failover twice a year without disrupting production. The application tier is managed by an instance pool and the database has a standby in Phoenix. The provider already has Resource Manager stacks that deploy every tier. Which design best satisfies the requirement to automate cross-region failover and recovery for the whole stack?
Full Stack Disaster Recovery coordinates the database role switch together with compute and load balancer recovery in a defined order, and it provides a rehearsal mode that does not disrupt production. A Data Guard standby plus daily boot volume copies cannot meet a 30-minute stack-wide RTO because restoring the compute tier is a manual, slow process. An always-running active copy doubles cost and does not provide the orchestrated, rehearsed failover requested, and a custom script leaves ordering and validation to the team.
A financial services firm is building an event-driven reconciliation workload. Each run is short-lived, executes a single container image, and must be triggered by an HTTP call from an internal application. The team wants no cluster to manage, pay-per-use billing, and the ability to run the task inside the company's private subnet with no public IP. An architect proposes OCI Container Instances. Which constraint in the requirements most directly determines whether Container Instances is the right choice over a Kubernetes-based design?
The decisive factors are the single short-lived container, the absence of cluster management, and pay-per-use billing, which together point to Container Instances rather than a Kubernetes cluster that carries standing node cost and operational overhead. Private subnet operation is supported by Container Instances and does not by itself favour it over OKE, while the trigger is an internal call so a public load balancer is not needed. The OCIR storage location is a registry choice that applies equally to both designs.
A payments company exposes a public API through an OCI Load Balancer with an OCI Web Application Firewall (WAF) policy. After a recent rule update, legitimate partner requests carrying a large JSON body began receiving 403 responses. The security team wants to keep the protection against injection attacks on all other endpoints and must not weaken the policy for the rest of the API. Which change best resolves the issue?
A scoped exclusion limits the relaxed behaviour to the specific path and parameter that triggers the false positive, preserving injection protection on every other endpoint. Detect-only mode removes blocking across the whole listener, removing the rule entirely weakens every endpoint, and bypassing the load balancer removes the WAF protection for partner traffic altogether.
Get all 100 questions & answers
Full access options for Oracle 1Z0-997-26
Self-study
PDF Version
- 100 actual exam questions
- Compatible with all devices
- Printable PDF format
- No download limits
- 90 days free updates
- 24/7 customer support
Best value
Bundle Pack
- Discounted PDF + online access
- 100 actual exam questions
- Both PDF and online practice test
- Free 90 days updates
- No download or practice limits
- 24/7 customer support
Exam simulation
Online Practice Test
- 100 actual exam questions
- Actual exam environment
- 90 days free updates
- Browser-based software
- 24/7 customer support
100% Money Back Guarantee
Your success comes first
At DumpsProvider, we give you a 100% money back guarantee on all of our exam dumps. For us, your success is the #1 priority and we are more than confident that our exam material will help you pass the 1Z0-997-26 Exam (Oracle Cloud Infrastructure Architect Professional) on your very first attempt.
In a rare case, if you fail your exam, we will refund your amount without asking a single question.