Database Management Systems

Tools are the fashion; Data Modeling is the skeleton. You can swap Airflow for Prefect, or Spark for DuckDB. But you can’t swap "bad logic" for a faster engine and expect it to work. In one project, I used Airflow. In another, Spark. Lately, it’s all dbt. But 100% of the time, the win came down to Data Modeling fundamentals. Building a data platform without modeling is like building a skyscraper on a swamp. It doesn't matter how expensive your gold-plated elevators (tools) are if the foundation is sinking. Here's what actually matters: 𝗗𝗶𝗺𝗲𝗻𝘀𝗶𝗼𝗻𝗮𝗹 𝗠𝗼𝗱𝗲𝗹𝗶𝗻𝗴 = 𝗦𝗽𝗲𝗲𝗱 Star schemas make queries fast. Facts and dimensions separated = happy analysts. 𝗦𝗖𝗗𝘀 𝗪𝗶𝗹𝗹 𝗕𝗶𝘁𝗲 𝗬𝗼𝘂 Skip SCD Type 2 tracking? Debug why historical reports show wrong data at 2 AM. 𝗡𝗼𝗿𝗺𝗮𝗹𝗶𝘇𝗮𝘁𝗶𝗼𝗻 𝗜𝘀𝗻'𝘁 𝗥𝗲𝗹𝗶𝗴𝗶𝗼𝗻 OLTP systems? Normalize for integrity. OLAP systems? Denormalize for speed. Know your world. Design accordingly. 𝗗𝗮𝘁𝗮 𝗩𝗮𝘂𝗹𝘁 = 𝗙𝗹𝗲𝘅𝗶𝗯𝗶𝗹𝗶𝘁𝘆 Business requirements changing weekly? Data Vault keeps you sane. Verbose but bulletproof. 👉 Here are the real Non-negotiables:  • Model for how data will be queried, not just stored  • Document your grain—ambiguity kills data trust  • Surrogate keys > natural keys (trust me on this)  • Test your model with real queries before building pipelines My 2 cents: Master data modeling, and every tool becomes easier. Skip it, and you'll spend your career firefighting broken pipelines. Are you willing to upskill❓Explore these resources: → Michael K.'s KahanDataSolutions - https://lnkd.in/g4JSFPph → Benjamin Rogojan's Seattle Data Guy - https://lnkd.in/ghewnvBX → The Data Warehouse Toolkit by Ralph Kimball - https://lnkd.in/dTynC6yD Image Credits: Shubham Srivastava Every pipeline you build will eventually be replaced. A solid data model? That becomes the language of the company. What's one data modeling mistake that cost you hours of debugging? Let's learn together. 👇

The wrong cloud data platform can become an expensive architectural decision. Databricks, Snowflake, Microsoft Fabric, Google BigQuery, and Amazon Redshift are all powerful but each is designed around a different operating model. Databricks is a strong fit for AI/ML, large-scale data engineering, streaming, and lakehouse workloads. Its biggest advantage is bringing data, analytics, and machine learning into one platform. Snowflake works well for enterprise data warehousing, high-concurrency SQL, governed data sharing, and elastic compute. It is especially attractive when simplicity and separation of storage and compute matter. Microsoft Fabric brings OneLake, Power BI, data engineering, and analytics together. It is a natural choice for organizations already invested in Microsoft 365, Azure, and the Power BI ecosystem. Google BigQuery is built for serverless analytics. It removes much of the infrastructure management and works well for organizations running large analytical workloads across Google Cloud. Amazon Redshift remains a strong option for AWS-first enterprises that need a mature MPP warehouse connected deeply with services such as S3, Glue, Kinesis, IAM, and SageMaker. The decision should not begin with feature lists. Start with these questions: ↳ Is your priority BI, AI/ML, streaming, or data engineering? ↳ Do you need a warehouse, lakehouse, or unified analytics platform? ↳ Which cloud ecosystem already runs your business? ↳ How important are open formats and portability? ↳ What pricing model matches your workload pattern? ↳ How much operational complexity can your team manage? There is no universally best data platform. The right choice is the one that fits your architecture, skills, governance requirements, workload behaviour, and long-term cloud strategy. Follow Ashish Joshi for more such insights!!

Most people learn SQL, Python, Spark, Azure, Databricks… But they skip data modelling. And then real job problems is, Reports don’t match. Business users question numbers. Pipelines become messy. One metric has 5 definitions. Nobody knows which table is the source of truth. I have seen this in real projects. Sometimes the problem is not the pipeline. Sometimes the problem is that the data was not modelled properly from the beginning. As a Data Engineer, you should know: → Fact vs Dimension tables → Star schema vs Snowflake schema → Slowly Changing Dimensions → Grain of a table → Primary keys and surrogate keys → Denormalization for analytics → How business metrics should be represented in tables This is what separates “I can move data” from “I can design useful data systems.” In interviews also, data modelling questions are becoming very common. They may ask: “How will you design a sales data model?” “How do you handle customer history?” “What is the grain of this fact table?” “How will you avoid duplicate revenue calculation?” Free resources to learn: 1. Kimball Dimensional Modelling https://lnkd.in/gWc2ZPMX 2. Microsoft Learn — Power BI Data Modelling https://lnkd.in/gtnFbH4b 3. Microsoft Fabric Dimensional Modelling https://lnkd.in/g8wYRRjz 4. dbt — Kimball Dimensional Model Tutorial https://lnkd.in/ghk_bepm 5. dbt Best Practices https://lnkd.in/gn3VChGi Don’t just learn tools. Learn how data should be structured. Because tools will change. But good modelling thinking will stay useful everywhere. Save this if you are preparing for Data Engineer interviews or building real projects. #DataEngineering #DataModeling #SQL #DataWarehouse #CareerGrowth

I just love how ❄️ Snowflake keeps making our lives easier 🙌 Snowflake has just released Performance Explorer to public preview, a visual way to monitor query workloads without writing SQL against ACCOUNT_USAGE, and here’s how it looks in one of our envs. 💡 What I found it helpful for: ⚆ Easily spot any type of anomaly in your environment ⚆ Track query duration, failures, and queuing issues ⚆ Get similar insights at the warehouse and table level ⚆ Compare metrics period-over-period to catch changes early The filters (warehouse/database/role) make it easy to drill down, and the side panels give you the details you need to dig deeper. Tip: Grant APP_USAGE_VIEWER and USAGE_VIEWER roles to your team so everyone can access it without needing ACCOUNTADMIN. Check it out under Monitoring >> Performance Explorer in Snowsight. 🔗 Official doc: https://lnkd.in/dBNSb9EF #snowflake #datasuperhero #snowflake_advocate #dataengineering Amilee Alesna

As we close out the year (and on the back of my 'Best of 2024' article yesterday), I'll post my top Data Ecosystem infographics and posts for the next few weeks. And let's start with Data Modelling! Data success in business does not start with an AI product, a well-constructed pipeline, or a frequently used dashboard. Success starts with the business model. This, then helps inform the org data model. Your 𝐝𝐚𝐭𝐚 𝐦𝐨𝐝𝐞𝐥 𝐰𝐢𝐥𝐥 𝐭𝐮𝐫𝐧 𝐭𝐡𝐞 𝐨𝐫𝐠𝐚𝐧𝐢𝐬𝐚𝐭𝐢𝐨𝐧’𝐬 𝐩𝐫𝐨𝐜𝐞𝐬𝐬𝐞𝐬 𝐢𝐧𝐭𝐨 𝐭𝐚𝐧𝐠𝐢𝐛𝐥𝐞 𝐝𝐚𝐭𝐚 𝐟𝐫𝐚𝐦𝐞𝐰𝐨𝐫𝐤𝐬 𝐚𝐧𝐝 𝐬𝐭𝐫𝐮𝐜𝐭𝐮𝐫𝐞𝐬 to drive business results. Unfortunately most organisations skip this step and get into a mess of data engineering quick fixes Here’s a breakdown of the process from business model to each part of the data model: 1. Start with the 𝐁𝐮𝐬𝐢𝐧𝐞𝐬𝐬 𝐌𝐨𝐝𝐞𝐥 🏢 Map out how your organisation creates, delivers, and captures value. For example, company creates product x, ships to store y, sells to customer z and makes money. Within that, what are the data points to track and understand? You map this out, you understand the process of how the organization makes money. 2. Extending it to the 𝐂𝐨𝐧𝐜𝐞𝐩𝐭𝐮𝐚𝐥 𝐃𝐚𝐭𝐚 𝐌𝐨𝐝𝐞𝐥 🗺️ The conceptual data model then builds on top of this, rewriting how the business operates using a high-level representation of organisational data. This phase involves identifying key data entities & domains and mapping the relationships between them. This should be understandable by business stakeholders as this is the bridge between business processes and data development. 3. Adding structure with the 𝐋𝐨𝐠𝐢𝐜𝐚𝐥 𝐃𝐚𝐭𝐚 𝐌𝐨𝐝𝐞𝐥 🏙️ Next, we refine into a logical data model. Here, we delve deeper to define the data types, attributes of each entity and the nature of the relationships. This model is still independent of technical implementation but sets out a clear structure (often through ERDs or UML) for how data is related and organised. 4. Building the 𝐏𝐡𝐲𝐬𝐢𝐜𝐚𝐥 𝐃𝐚𝐭𝐚 𝐌𝐨𝐝𝐞𝐥 foundations 🏗️ Finally, we arrive at the physical data model. It’s the actual database schema design, complete with tables, columns, data types, and constraints. It also takes into account the performance requirements, optimization techniques, and the physical storage of the data. Obviously this is a broad simplification of the process. Each step of this journey requires collaboration between business leaders, data architects, and engineers to ensure the data models align with the strategic goals, is technically feasible and is understood by all relevant stakeholders. Building this properly helps everybody understand how data actually delivers value. Too often we engineer without the architecture to structure it. Fix this and you will be way better off in the long-term. #DataModeling #DataEngineering #BusinessModel #DataArchitecture #DataStrategy #DylanDecodes

𝗘𝘃𝗲𝗿 𝗯𝗲𝗲𝗻 𝗮𝘀𝗸𝗲𝗱 𝗮 𝗱𝗮𝘁𝗮 𝗺𝗼𝗱𝗲𝗹𝗶𝗻𝗴 𝗾𝘂𝗲𝘀𝘁𝗶𝗼𝗻 𝗶𝗻 𝗮𝗻 𝗶𝗻𝘁𝗲𝗿𝘃𝗶𝗲𝘄 — 𝗹𝗶𝗸𝗲 𝗱𝗲𝘀𝗶𝗴𝗻𝗶𝗻𝗴 𝗮 𝗰𝘂𝘀𝘁𝗼𝗺𝗲𝗿–𝗽𝗿𝗼𝗱𝘂𝗰𝘁–𝗼𝗿𝗱𝗲𝗿 𝘀𝗰𝗵𝗲𝗺𝗮 — 𝗮𝗻𝗱 𝘀𝘂𝗱𝗱𝗲𝗻𝗹𝘆 𝗳𝗲𝗹𝘁 𝘀𝘁𝘂𝗰𝗸? You’re not alone. This happens even to experienced data engineers who spend most of their time building pipelines, tuning Spark jobs, or managing cloud infra. Here’s the hard truth I’ve learned over time: 𝗡𝗼 𝗺𝗮𝘁𝘁𝗲𝗿 𝗵𝗼𝘄 𝗺𝗼𝗱𝗲𝗿𝗻 𝘆𝗼𝘂𝗿 𝘀𝘁𝗮𝗰𝗸 𝗶𝘀 — 𝗔𝘇𝘂𝗿𝗲, 𝗗𝗮𝘁𝗮𝗯𝗿𝗶𝗰𝗸𝘀, 𝗦𝗻𝗼𝘄𝗳𝗹𝗮𝗸𝗲, 𝗣𝗼𝘄𝗲𝗿 𝗕𝗜 — 𝗶𝗳 𝘆𝗼𝘂𝗿 𝗱𝗮𝘁𝗮 𝗺𝗼𝗱𝗲𝗹 𝗶𝘀 𝘄𝗲𝗮𝗸, 𝘆𝗼𝘂𝗿 𝗲𝗻𝘁𝗶𝗿𝗲 𝘀𝘆𝘀𝘁𝗲𝗺 𝗶𝘀 𝗳𝗿𝗮𝗴𝗶𝗹𝗲. After multiple interviews and real-world projects, one thing became very clear to me. The strength of a data platform comes down to 𝘁𝘄𝗼 𝗳𝘂𝗻𝗱𝗮𝗺𝗲𝗻𝘁𝗮𝗹𝘀 👇 1️⃣ 𝗦𝘁𝗿𝗼𝗻𝗴 𝗗𝗮𝘁𝗮 𝗠𝗼𝗱𝗲𝗹𝗶𝗻𝗴 Data modeling isn’t about drawing boxes in an ER diagram. It’s about answering 𝗱𝗲𝘀𝗶𝗴𝗻-𝗱𝗲𝗳𝗶𝗻𝗶𝗻𝗴 questions: • Should this be a 𝗳𝗮𝗰𝘁 or a 𝗱𝗶𝗺𝗲𝗻𝘀𝗶𝗼𝗻? • What’s the 𝗴𝗿𝗮𝗻𝘂𝗹𝗮𝗿𝗶𝘁𝘆 — transaction-level, daily, aggregated? • Are we using 𝗻𝗮𝘁𝘂𝗿𝗮𝗹 𝗸𝗲𝘆𝘀 𝗼𝗿 𝘀𝘂𝗿𝗿𝗼𝗴𝗮𝘁𝗲 𝗸𝗲𝘆𝘀, and why? • How do we handle 𝗦𝗹𝗼𝘄𝗹𝘆 𝗖𝗵𝗮𝗻𝗴𝗶𝗻𝗴 𝗗𝗶𝗺𝗲𝗻𝘀𝗶𝗼𝗻𝘀 (𝗦𝗖𝗗𝘀) without breaking history? • Are we trading off 𝗱𝗮𝘁𝗮 𝗿𝗲𝗱𝘂𝗻𝗱𝗮𝗻𝗰𝘆 𝘃𝘀 𝗿𝗲𝗳𝗲𝗿𝗲𝗻𝘁𝗶𝗮𝗹 𝗶𝗻𝘁𝗲𝗴𝗿𝗶𝘁𝘆 correctly? A well-designed dimensional model: • Simplifies analytics • Enables true self-service BI • Reduces query complexity • Prevents silent data inconsistencies 2️⃣ 𝗦𝗼𝗹𝗶𝗱 𝗗𝗮𝘁𝗮 𝗪𝗮𝗿𝗲𝗵𝗼𝘂𝘀𝗶𝗻𝗴 𝗣𝗿𝗶𝗻𝗰𝗶𝗽𝗹𝗲𝘀 Even in lakehouse or medallion architectures, warehouse thinking still matters. • Bronze → Silver → Gold layers enforce 𝗱𝗮𝘁𝗮 𝗾𝘂𝗮𝗹𝗶𝘁𝘆, 𝗹𝗶𝗻𝗲𝗮𝗴𝗲, 𝗮𝗻𝗱 𝘁𝗿𝘂𝘀𝘁 • Partitioning & clustering decide whether queries scale or crawl • Star vs Snowflake schema impacts 𝗣𝗼𝘄𝗲𝗿 𝗕𝗜 𝗽𝗲𝗿𝗳𝗼𝗿𝗺𝗮𝗻𝗰𝗲 𝗮𝗻𝗱 𝗗𝗔𝗫 𝗰𝗼𝗺𝗽𝗹𝗲𝘅𝗶𝘁𝘆 • Denormalization improves read performance — but increases maintenance cost I’ve seen teams spend months building pipelines, only to realize dashboards are slow because: 𝗳𝗶𝗹𝘁𝗲𝗿𝘀 𝗱𝗼𝗻’𝘁 𝘀𝗹𝗶𝗰𝗲 𝗳𝗮𝗰𝘁 𝘁𝗮𝗯𝗹𝗲𝘀 𝗰𝗼𝗿𝗿𝗲𝗰𝘁𝗹𝘆 — 𝗮 𝗽𝘂𝗿𝗲 𝗺𝗼𝗱𝗲𝗹𝗶𝗻𝗴 𝗶𝘀𝘀𝘂𝗲. 𝗙𝗶𝗻𝗮𝗹 𝗧𝗵𝗼𝘂𝗴𝗵𝘁 Whether you’re preparing for interviews or already working as a data engineer: 𝗗𝗮𝘁𝗮 𝗺𝗼𝗱𝗲𝗹𝗶𝗻𝗴 𝗶𝘀𝗻’𝘁 𝗼𝗽𝘁𝗶𝗼𝗻𝗮𝗹. 𝗜𝘁’𝘀 𝘁𝗵𝗲 𝗳𝗼𝘂𝗻𝗱𝗮𝘁𝗶𝗼𝗻 𝗲𝘃𝗲𝗿𝘆𝘁𝗵𝗶𝗻𝗴 𝗲𝗹𝘀𝗲 𝘀𝘁𝗮𝗻𝗱𝘀 𝗼𝗻. Pipelines can be rewritten. Infrastructure can be scaled. A broken data model is much harder to fix. Curious to know 👇 𝗛𝗮𝘃𝗲 𝘆𝗼𝘂 𝗯𝗲𝗲𝗻 𝗮𝘀𝗸𝗲𝗱 𝗱𝗮𝘁𝗮 𝗺𝗼𝗱𝗲𝗹𝗶𝗻𝗴 𝗾𝘂𝗲𝘀𝘁𝗶𝗼𝗻𝘀 𝗶𝗻 𝗶𝗻𝘁𝗲𝗿𝘃𝗶𝗲𝘄𝘀? 𝗢𝗿 𝗹𝗲𝗮𝗿𝗻𝗲𝗱 𝘁𝗵𝗶𝘀 𝗹𝗲𝘀𝘀𝗼𝗻 𝘁𝗵𝗲 𝗵𝗮𝗿𝗱 𝘄𝗮𝘆 𝗼𝗻 𝗮 𝗽𝗿𝗼𝗷𝗲𝗰𝘁? Follow Aishwarya Pani for more practical, real-world data engineering insights.

Difference between Data Object Graphs and Knowledge Graphs Lot's of people have asked me how Data Object Graphs (DOGs) differ from traditional Knowledge Graphs (KGs). So I thought I share my perspectives: The Key Difference (TL;DR): While Knowledge Graphs excel at representing what things ARE, Data Object Graphs excel at modelling (and executing) what things DO. Specifically, Data Object Graphs (DOGs): - Model direct behaviour and are direct business process implementations - DOGs map 1:1 to actual business operations, enabling rapid translation from modelling to execution - Represent and are the exact business process come to life (e.g., Customer → appliesForLoan → RiskDept → scoresLoan → ReportCreator →createsReport) - Include specific executable steps that are implemented by a specific executable data product containers (e.g. Metrics, ML models, decisions etc..) with interactions directly mapped to business actions (specific API / graph calls). - Have agility - Business changes can be quickly implemented without complex intermediate abstractions - Provide immediate operational value through executable modelling - Faithfully model the real-world - ie what actually happens in your business, not theoretical abstractions - Enables simulation of business ideas with real data and execution capabilities Knowledge Graphs (KGs): - Tend to focus on formal semantic relationships between conceptual entities - Emphasize standardized ontologies and taxonomies - Excel at representing knowledge relationships beyond operational contexts - Provide semantic reasoning capabilities through formalized structures - Designed primarily for knowledge representation rather than process execution The DOG approach allows organizations to model during business process development rather than at an abstract data level, solving brittleness problems while maintaining enterprise-wide connectivity. The DOGs can go from business process modelling directly to execution with remarkable speed and agility. This allows organizations to adapt quickly to changing business requirements without sacrificing enterprise-wide visibility. These are two very valuable approaches, but have different objectives and goals but can be complementary #DataArchitecture #KnowledgeGraphs #DataObjectGraphs #BusinessProcessModeling #EnterpriseAgility #DataProducts

Choosing the Right Database Made Simple As data engineers, picking the right database can be tricky. It’s not just about storing data—it’s about understanding its journey. Here's a quick guide: 🔴 Data Flow Patterns Heavy Writes: Use Apache Cassandra or TimescaleDB for time-series data. Read-Heavy Apps: Go for Redis or MongoDB with read replicas. ACID Compliance: PostgreSQL or MySQL are your go-to options. 🔴 Scaling Needs Horizontal Scaling: Choose DynamoDB or Cassandra for distributed systems. Vertical Scaling: PostgreSQL works well for single powerful instances. Global Reach: CockroachDB or Azure Cosmos DB for multi-region setups. 🔴 Data Complexity Complex Relationships: Neo4j for graph-based data. Document Storage: MongoDB or CouchDB for flexible schemas. Time-Series Data: InfluxDB or TimescaleDB. Search-Heavy Apps: Elasticsearch for full-text search. 🔴 Operational Overhead Managed Services: Cloud options like RDS or Atlas for less maintenance. Self-Hosted: Choose based on team expertise. Backup & Recovery: Check for replication and recovery features. 🔴 Performance Query Patterns: Optimize for frequent queries. Indexing: Ensure efficient indexing. Memory vs. Disk: Use Redis for ultra-low latency. 🔴 Costs Storage Growth: Plan for scaling expenses. Query Costs: Monitor costs in cloud-based solutions. Operational Costs: Include monitoring and maintenance. Real-World Examples: User Tracking: Cassandra (high write throughput). Financial Transactions: PostgreSQL (ACID compliance). Content Management: MongoDB (flexible schema). Real-Time Analytics: ClickHouse (fast aggregations). Cache: Redis (in-memory, fast). Pro Tip: Start with a proven solution like PostgreSQL unless you need something specific. Scaling a reliable system is easier than fixing an exotic one in production. Cloud Database Options: AWS: DynamoDB, ElastiCache, Redshift. Google Cloud: BigQuery, Firestore, Cloud SQL. Azure: Cosmos DB, Redis Cache, Data Lake Storage. CC:Rocky Bhatia #Data #Engineering #SQL #Databases

SQL vs. NoSQL: Cheatsheet for AWS, Azure, and Google Cloud This cheat sheet outlines the major types of SQL and NoSQL databases, their use cases, and their corresponding implementations across AWS, Azure, Google Cloud, and cloud-agnostic solutions. ➥ Structured Data  1. Relational (ACID Transactions, OLTP)   Use Case: Transactional systems requiring consistency (e.g., banking, ERP).   - AWS: RDS, Aurora   - Azure: Azure SQL Database   - Google Cloud: Cloud SQL, Cloud Spanner   - Cloud Agnostic: SQL Server, Oracle, DB2, MySQL, PostgreSQL   2. Columnar (Analytics, OLAP)   Use Case: Analytics, reporting, large-scale aggregation.   - AWS: Redshift   - Azure: Azure Synapse   - Google Cloud: BigQuery   - Cloud Agnostic: Snowflake, ClickHouse, Druid, Pinot, Databricks  ➥ Semi-Structured Data  3. Key-Value (Dictionary, Cache)   Use Case: Fast access to small data payloads, caching.   - AWS: DynamoDB, ElastiCache   - Azure: Cosmos DB, Azure Cache for Redis   - Google Cloud: BigTable, Memorystore   - Cloud Agnostic: Redis, Memcached, Hazelcast, Ignite   4. Wide Column (2-D Key-Value)   Use Case: Handling semi-structured data at scale.   - AWS: Keyspaces   - Azure: Cosmos DB   - Google Cloud: BigTable   - Cloud Agnostic: HBase, Cassandra, ScyllaDB   5. Time Series   Use Case: Monitoring, time-based data like IoT metrics.   - AWS: Timestream   - Azure: Cosmos DB   - Google Cloud: BigTable, BigQuery   - Cloud Agnostic: OpenTSDB, InfluxDB, ScyllaDB   6. Immutable Ledger (Audit Trail)   Use Case: Storing immutable records for compliance and auditing.   - AWS: Quantum Ledger Database (QLDB)   - Azure: Azure SQL Database Ledger   - Google Cloud: Not Applicable   - Cloud Agnostic: Hyperledger Fabric   7. Geospatial (Location & Geo-entities)   Use Case: Geographic data storage and processing.   - AWS: Keyspaces   - Azure: Cosmos DB   - Google Cloud: BigTable, BigQuery   - Cloud Agnostic: Solr, PostGIS, MongoDB (GeoJSON)   8. Graph (Entity-Relationships)   Use Case: Relationship-centric queries, social networks, and recommendation engines.   - AWS: Neptune   - Azure: Cosmos DB   - Google Cloud: JanusGraph + BigTable   - Cloud Agnostic: OrientDB, Neo4J, Giraph   9. Document (Nested Objects: XML, JSON)   Use Case: Storing hierarchical data structures.   - AWS: Document DB   - Azure: Cosmos DB   - Google Cloud: Firestore   - Cloud Agnostic: MongoDB, Couchbase, Solr   10. Text Search (Full-Text Search)   Use Case: Search systems for large datasets.   - AWS: OpenSearch, CloudSearch   - Azure: Cognitive Search   - Google Cloud: Search APIs on Datastores   - Cloud Agnostic: Elasticsearch, Solr, Atlas  ➥ Unstructured Data  11. (Rich Text, Images, Videos)   Use Case: Storage for unstructured content like images, videos, and documents.   - AWS: S3   - Azure: Blob Storage   - Google Cloud: Cloud Storage   - Cloud Agnostic: HDFS, MinIO