AZURE project questions
- "Tell me about an Azure project you've worked on."
"In my last role, we needed to migrate a legacy, on-premises e-commerce platform to Azure because our local data centre couldn't handle sudden traffic spikes and scaling hardware was too slow and expensive
- To solve this, I designed and implemented a fully cloud-native architecture. We containerised the application microservices using Docker and deployed them onto Azure Kubernetes Service (AKS) for seamless, automated scaling. For the data layer, we migrated our relational database to Azure SQL Database using the Azure Database Migration Service to ensure minimal downtime. We also integrated Azure Blob Storage to host static assets and media files, which significantly reduced the load on our primary compute instances. [1, 2, 3]My specific responsibility was setting up the secure networking and deployment pipelines. I configured Azure Virtual Networks (VNets), restricted access via Network Security Groups (NSGs), and managed all application secrets securely using Azure Key Vault. I also built the Azure DevOps CI/CD pipelines to automate our infrastructure deployments using Terraform. [1, 2]As a result of this migration, we eliminated application downtime during high-traffic sales events. Our deployment cycles dropped from once a week to multiple times a day, and the company reduced overall infrastructure hosting costs by 25% due to Azure's pay-as-you-go model." [1]Project Title: On-Premises Insurance Claims Document Migration to Azure StorageProject OverviewThis project involved the cloud migration of legacy insurance claims documentation from an on-premises infrastructure to a scalable cloud storage solution.
- The Nature of the Data: The dataset consisted entirely of unstructured data, specifically digitized insurance claim files. These documents contained scanned variants of physical paper claim forms, handwritten accident descriptions, binding customer signatures, and embedded high-resolution photographic evidence of vehicle or property damage.
- The Business Rationale for PDF Format: The insurance industry relies heavily on these documents for legal compliance, audit trails, and dispute resolution. The PDF format functions as a digital photocopy. It permanently preserves the exact layout, visual signatures, and contextual imagery within a single, unalterable, and universally readable file format.
- The Technical Rationale against Relational Architecture: Relational database systems (such as Microsoft SQL Server) are engineered exclusively for structured data organized into strict tables, columns, and data types (e.g., integers, strings, or dates). Complex components like physical signatures, variable handwritten text, and multi-megabyte image binaries do not fit into a traditional tabular schema. Attempting to store them as Large Objects (BLOBs) inside a relational system degrades database performance and inflates computing costs. Consequently, unstructured files require dedicated cloud object storage rather than a relational engine.
1. What Was the Existing Architecture- Compute & Storage: A single, localized Windows desktop machine configured to function as a basic Network-Attached Storage (NAS) file share within the local office area network (LAN).
- Ingestion: Insurance adjusters manually scanned physical documents and saved the resulting PDF files directly into specific shared folders on this localized machine.
- Disaster Recovery: Redundancy was managed via a manual process where an employee copied the updated files onto an external USB hard drive once every Friday afternoon.
2. What Issues Was He Facing with Existing Architecture- Storage Exhaustion: The local machine’s hard drive had reached 98% capacity due to the high volume of incoming scanned PDFs, completely halting the ingestion of new insurance claims.
- Operational Silos: Field agents conducting on-site accident or property assessments could not access historical claim files or previous documentation while outside the physical office network.
- Single Point of Failure (SPOF): The business faced severe data loss risks. A localized hardware failure, power surge, ransomware attack, or physical office disaster would result in the permanent destruction of historical customer records.
3. How Did You Capture the Requirement List- Capacity Planning: I executed a storage analysis script on the local file share to determine the exact data footprint, which measured 350 Gigabytes (GB) across thousands of individual PDF files.
- Security & Compliance Auditing: I consulted with the risk and compliance manager to establish data governance requirements. The solution demanded data encryption at rest and restricted, audited access control to protect sensitive personally identifiable information (PII).
- Bandwidth Assessment: I performed a network throughput and latency test on the office internet connection to calculate the time window required to transfer 350 GB of data without disrupting business operations.
4. How Did You Design the Solution to the Client- Target Cloud Architecture: I selected Azure Blob Storage (Hot Tier). This object storage service provides cost-effective storage for unstructured file data, built-in high availability, and seamless scalability without infrastructure overhead.
- Logical Organization: I provisioned a standard Azure Storage Account and established a dedicated Blob Container structured as
insurance-claims-pdf. - Security Framework: I enforced Azure Storage Encryption using Microsoft-managed keys for data-at-rest protection. To secure access, I configured Shared Access Signatures (SAS tokens) with short-term expiration windows and specific read/write permissions, ensuring only validated personnel could interact with the data.
- Migration Strategy: I deployed Azure Storage Explorer on the local server. Using this tool, I executed a data migration wave over a single weekend maintenance window, successfully transferring the entire 350 GB dataset to Azure without causing operational downtime.
No comments:
Post a Comment