What every Azure Developer ought to Know?

Microsoft Azure has been widely adopted especially since its hybrid cloud features. There are 100s of services available which kind of make developers and even architects quite confusing which one to choose. I ‘ll try to be concise about them and compare them in terms of features (not in terms of pricing) so you can focus only those which matters to your project needs.

There are many services which enable you to run your applications in Azure. I ‘ll divide them into three categories;

Infrastructure-as-a-service (IaaS)

VM
Container Service
Container Instance
Web App for Container
Batch
Service Fabric

Platform-as-a-service (PaaS)

Cloud Services
Mobile App
Web App

Logic-as-a-service (LaaS)

Function App
Logic App

Few notable bits about aforementioned services:

Virtual Machines (VM)

Start and stop in minutes
Many images (Win/Linux) to choose from, you can import your own image or even bring your own licensed version
Many sizes to choose from based on CPU/GPU, Memory, Network Bandwidth, Storage
SLA 99.95% (provided you are running multiple instances)

Azure Container Service

Allow multiple containers
Unlike VM, containers start/stop in seconds
Comes with containers orchestrator (Kubernetes, Docker Swarm, etc)
Run Linux and Windows-based containers
Use images from Docker Hub or Azure container Registry

Azure Container Instance

Run single container
Start/stop in seconds
No orchestrator
Run Linux and Windows-based containers
Use images from Docker Hub or Azure container Registry

Azure Web App Container

Allow container on App Service
No orchestrator needed
Only Linux based container
App service features

Easy continuous deployments
Deployment slots
Auto Scaling

Azure Batch

Run jobs across multiple VM or Containers (if needed)
Use for intensive jobs
Orchestrator managing jobs
Run jobs on Windows/Linux

Azure Service Fabric

Allow orchestrator to run web app containers
Allow executable
Allow stateful/stateless services
Allow on-premise deployment
Allow App service features

Azure Cloud Services

Run your app in VM as a web or worker role
You can control VM via RDP access if you want (but not necessary)
Provides scaling
An app could be web-based or even background job

Azure App Services (Web App)

Runs as a web server (Apache, IIS, Tomcat, etc)
Allow a range of different fx based applications (.net, python, nodejs)
App service features (CD, Deployment slots, custom Domain, autoscaling, auth, web jobs)
Allow on-premise resource access
SLA 99.95%

Azure Web Services (Web Job)

Run inside Web App
Interval-based or external trigger based like Message on Q

Run inside Web App
Interval-based or external trigger based like Message on Q

Azure Mobile App

Runs your mobile app as background service
Allow any mobile SDK (ios, android, etc)
Unique features
Offline sync
Push notifications

Function Apps

Deploy and execute only code
Can be triggered based on interval or event (message on Q)
Have Binding
Auto Scale
Pricing based on consumption only
Allow a range of technologies (.net, nodejs)

Logic Apps

Execute pipeline of tasks
Biztalk services replacement
Can be triggered
Can connect external connectors like office365, Twitter, Custom API
Auto Scale

Following tables should make your life easier to decide:

Comparing VM/Container based services

Comparing API/Background Task-based services

Where to run your App

Azure Security Services

It’s super important to assure your application’s security. Azure provides security services that make your life easy in securing your applications. Different services provide different security-related features, Authentication, Vault, Auto encryption database, etc.

I ‘ll briefly describe them below, in terms of what you should choose and when:

Azure Active Directory

Identity Provider-as-a-service manage authentication information
Store Identities — like user information email, username, etc
Authentication & Authorization (login, password reset, sign up pages)
Integrate with on-premise IDPs
Protect applications on-premise and on cloud
Multi-factor authentication
Advanced threat detection which detects malicious behavior like if you are logged in from one region and second request comes from totally different geo-location, it will ask for 2-factor authentication

Azure Key Vault

Store and encrypt keys (certificates)
Store secrets (passwords, connection strings, hash)
Developer Read/Write via API/SDK doesn’t have to store sensitive data into the source code
Auto logging regarding secret data operations

Azure App service certificates

Provision TLS certificates
Buy certificates from Azure directly
Simplifies domain verification
Auto-renewal
Store certificates in Azure key vault
Use certificates on App services, Azure app, on-premise app

Data Store Encryption

Encrypt data at rest
Store data in encrypted format automatically
Encrypt and Decrypts automatically
No noticeable performance overhead

Comparison

Azure Storage Services

Data storages are classifieds into multiple categories, First I ‘ll divide them in two major categories e.g, OLAP and OLTP

OLTP — Online Transactional Processing

Relational

Azure SQL DB
Azure MySQL
Azure PostgreSQL
Azure Maria DB

Non Relational — Unstructured/Document data (NoSql)

Azure Cosmos DB

Azure Storage

File
BLOB
Table
Queue
Disk

OLAP — Online Analytical Processing

Azure SQL Data Warehouse
Azure Data Lake Store

Following are some bits about the aforementioned Azure storage options:

Azure SQL Database

It’s just MS SQL Server in the cloud
Managed and Scalable
99.99% Availability and redundant (backups every hour to every 5 minutes)
Accessible via SSMS, Visual Studio, Azure Portal
Geo-Replication (single writable node other read only)
Automatic Tuning
Automatic Thread Detection
Dynamic Data masking for sensitive data like Credit Card data

Azure MySQL

MySQL in the cloud
Managed and Scalable
99.99% Availability and redundant (backups every hour to every 5 minutes)
Accessible via MySQL workbench and SQLyog

Azure PostgreSQL

PostgreSQL in the cloud
Managed and Scalable
99.99% Availability and redundant (backups every hour to every 5 minutes)
Accessible via pg-admin

Azure Maria Db

MariaDB in the cloud
Managed and Scalable
99.99% Availability and redundant (backups every hour to every 5 minutes)
Accessible via MySQL workbench and SQLyog
Support many storage engines

Apache Cassandra
CSV
Column Store

Azure CosmoDB

Distributed massive scale DB
No Schema
Auto Indexing
GeoReplication (single writable node)
SLA 99.99%
Low latency from Azure cloud < 10ms for reading and <15ms for write
Use it with several APIs

Document Db API
MongoDB API
Table API
Graph API
Apache Cassandra API

Azure Storage

Massive scale
Multi-purpose storage

File Storage (can mount as a File share)
BLOB Storage (allow multiple blobs like Block/Page and Append)
Q Storage
Table Storage (semi-structured data)
Disk Storage (high performance, can port Apps, perfect for VMs)

Shared access signature
Firewall and VPN
Data encryption at rest (just enable it)
Storage replication

Locally
Geo-Replica
Read-access Geo-Replica

Azure SQL Data Warehouse

Structured and Relational
Optimized for reporting and BI
A large amount of data (like 1PB)
Data encryption at rest

Azure Data Lake Store

Structured and Non Structured
Allow data in native format
No schema needed
Optimized for reporting
No size limit
Data encryption at rest

Azure Performance Services

The good thing about this section is, there are very few but very effective services to discuss

1. Azure CDN (Content Delivery Network)

Suitable for Static content like js/CSS/HTML/media files
Geographically data is accessible provides low latency

2. Azure Redis Cache

Open source Redis Cache
Store KeyValue Pair (hashes and lists)
Geo-Replication
Data Persistence
Much faster than any database as data coming from memory

Azure Traffic Manager

Routing service
Help to Load balancing (Priority-Weight)
Periodically ping for Health checks
Not bound to Geolocation
Geo-Performance

Comparison

Azure Message Queues Services

Azure Storage Queue

Simple and Basic
Allow only one receiver
Allow only text body
Size limit 64KB
Transaction Logs

Azure Service Bus Queue

FIFO ordering
One receiver for each message
Message consists on Key/Value pair (for additional information)
Body can be binary, json, text)
Message size limit upto 1MB
Duplicate detection
Dead letter Q… for corrupted or failed message, message automatically move to this Q so you can later debug them

Azure Bus Service Topic

Message can be processed by multiple receivers
Message filter (for different type of messages different receiver can act)
Message consists on Key/Value pair (for additional information)
Body can be binary, json, text)
Message size limit upto 1MB
Duplicate detection
Dead letter Q… for corrupted or failed message, message automatically move to this Q so you can later debug them

Azure Event Grid

Use Subscriber to and push events
Can subscribe to many events
Push to any destination (no pulling, retry if failed)
Pay as you go

Azure Event Hub

Use to ingest massive amount of data
Allow multiple sender and receiver
Message consists on Key/Value pair (for additional information)
Body can be binary, json, text)
Message size limit upto 256KB
Replay events
Retention between 1–7 days
Event data archive (in BLOB storage)

Azure IOT hub

Use to ingest massive amount of data
Allow multiple sender and receiver
Message consists on Key/Value pair (for additional information)
Body can be binary, json, text)
Message size limit upto 256KB
Manage Security devices
Bidirectional (device to cloud and vice versa)

Azure Notification Hubs

Use it to push to many subscribers
Platform Notification Services (PNS) — Androind, iOS, Kindle etc
Easily integrate with Azure mobile apps
Workflow
Register device PNS handle
Register PNS handle with Hub
Send notification to hub and hub forward to subscribers

Comparison

Azure Monitoring Services Comparison

TLDR;

There are an excess of 100 Azure services which I doubt can be aced by anybody however this article should give pretty good idea to developers and development architects which one they should concentrate based on their application needs.

27 Oct 2020

Software Development