Type of data that is to be stored
Structured
semi-structured
unstructured
The factors depend on operational needs
1. How often is the data accessible?
2. How quickly do we need to serve?
3. Need to run simple Queries?
4. Need to run complex queries?
5. accessed from Multiple regions?
1. Azure Storage Account: Think of an Azure Storage Account as the top-level resource that groups together various Azure Storage services. When you create a "General-purpose v2" storage account, it can contain:
Azure Blob Storage: For unstructured data (files, images, videos, backups).
Azure File Storage: For managed file shares (SMB/NFS).
Azure Queue Storage: For messaging (queues).
Azure Table Storage: For NoSQL key-value data.
Azure Disks: For VM disks (though often managed separately as "Managed Disks").
2. Azure Blob Storage:
This service is for storing unstructured data (Binary Large Objects or "blobs").
It's like a vast digital bucket where you put files. These files are accessed via HTTP/HTTPS endpoints.
It has different "types of blobs" (Block, Page, Append) and "access tiers" (Hot, Cool, Archive) as we discussed, but none of these are "queues."
3. Azure Queue Storage (The Actual Queue Service):
Purpose: Azure Queue Storage is a service specifically designed for storing large numbers of messages.
Its primary use case is to decouple components of an application, enabling asynchronous communication. How it works:
One part of an application (the "producer") places messages into a queue.
Another part of the application (the "consumer" or "worker") retrieves and processes these messages at its own pace.
This makes applications more scalable, resilient, and responsive.
Key Characteristics:
Messages: Each message can be up to 64 KB in size.
Retention: Messages can remain in the queue for up to 7 days by default (and can be set to never expire for newer API versions).
Access: Messages are typically accessed on a First-In, First-Out (FIFO) basis, though due to its distributed nature, strict order isn't guaranteed without specific design patterns.
Visibility Timeout: When a message is retrieved, it becomes "invisible" to other consumers for a configurable period (visibility timeout), preventing multiple workers from processing the same message simultaneously.
If processing fails, the message reappears after the timeout. Scalability: Can store millions of messages, up to the total capacity limit of the storage account.
Azure Data Lake Storage Gen2 (ADLS Gen2): This is Microsoft's enterprise-grade, highly scalable, and cost-effective data lake solution built on top of Azure Blob Storage. Its key differentiator is the Hierarchical Namespace (HNS), which allows it to organize data into folders and subfolders, much like a traditional file system. This HNS is critical for the performance and compatibility with big data analytics engines like Apache Spark and Hadoop, which expect a file system-like interaction.
Delta Lake: This is an open-source storage layer that brings ACID transactions (Atomicity, Consistency, Isolation, Durability) to data lakes. It was originally developed by Databricks and operates on top of existing cloud object storage (like ADLS Gen2). Delta Lake fundamentally transforms a traditional data lake into a "Lakehouse" architecture, combining the flexibility and low cost of a data lake with the reliability and performance typically found in data warehouses.
What is Azure Delta Lake Gen2?
When we talk about "Azure Delta Lake Gen2," we're referring to implementing Delta Lake as the storage format and transaction layer directly on files stored in Azure Data Lake Storage Gen2.
Essentially:
ADLS Gen2 provides the foundational, massively scalable, and cost-effective storage for your raw data files (typically Parquet, ORC, JSON, CSV).
Delta Lake sits on top of these files, adding a transactional layer (via a transaction log) and metadata management that provides features traditionally associated with databases.
Comments
Post a Comment