Cloud computing is, Compute, Networking, Storage and other services - managed by someone else. That can be rented. And returned.

1. 1. You buy the hardware
2. 2. You buy the softwares
3. 3. You employ an IT team to manage the hardwares
4. 4. All this costs money. And the money, is not OPEX but CAPEX.
5. 5. And, of course, as the software and hardware age - they need to be replaced.

Cloud computing is, Compute, Networking, Storage and other services - managed by someone else. That can be rented. And returned.

1. 1. You rent the hardware
2. 2. You buy the softwares
3. 3. You employ few specialists to manage the infrastructure.
4. 4. All this costs money. And the money, is OPEX and not CAPEX. You pay only for what you use.
5. 5. And, of course, as the software and hardware age. But this time, the software can be upgraded. And, the harware rental can be swapped for newer ones.

Characteristics:

1. 1. On-demand self service

   No human interaction is needed for resource provisioning. Available 24/7.

2. 2. Broad Network Access

   Create a network in one part of the world and access it from another. Access is not physical.

3. 3. Resource pooling

   Resource is being shared across, multiple customers. But, can be limited.

4. 4. Rapid Elasticity

   Can be scaled up and down as needed. Within minutes.

5. 5. Measured Service

   Payment made for resource used. No additional costs. Measurement: Server time or DB Time or Function calls / unit time.

1. 1. IAAS - Infrastructure As A Service
2. 2. PAAS - Platform As A Service
3. 3. SAAS - Software As A Service

1. 1. FAAS - Functions As A Service
2. 2. DbAAS - Database As A Service
3. 3. DAAS - Desktop As A Service
4. 4. IOTAAS - IOT As A Service
5. 5. AIAAS - AI As A Service

1. 1. Public cloud
2. 2. Hybrid cloud
3. 3. Private cloud

Regions: Each data centers location is called a region. There are ~60 regions.

Americas	Europe	Africa	Asia Pacific
Brazil South	France Central	South Africa North	Australia East
Canada Central	Germany West Central		Central India
Central US	North Europe		Japan East
East US	Norway East		Korea Central
East US 2	UK South		Southeast Asia
South Central US	West Europe		East Asia
US Gov Virginia	Sweden Central
West US 2
West US 3

Availability zones: A datacenter is called a zone. When there are more than one data center in a region, its called an Availability zone.

Subscription: Everything you subscribe inside the Azure portal.

Account: Your Azure account (identity) using which you can access your subscriptions..

• 1. Menu -> Resource groups -> Create
• 2. Choose a Subscription
• 3. Enter a resource group name
• 4. Sepcify a region.
• 5. Review & Create -> Create

Doing the same with a command line: az group create -l westus -n CLITest-rg

Doing the same with a PowerShell: New-AzResourceGroup -Name PSTest-rg Location westus

Heirarchy of ownership:

• 1. Management groups
• 2. Subscriptions
• 3. Resource groups
• 4. Resources.

Storage accounts: Costs are extemely cheap.

SLA: Costs are extemely cheap. Different SLA's offer 99.9, 99.99, 99.99% uptimes. Free and shared resources don't have SLA's.

Pricing models:

• 1. Per resource
• 2. Per consumption
• 3. Reservations
• 4. Resources.

• 1. Set of services for hosting and running applications
• 2. Allows uploading code and executing it
• 3. Offers various levels of control and flexibility.

Azure - Virtual machines

• A virtual machine is a software, which runs on an OS like any other software. But this is capable of emulating an OS.
• Allows creating new servers quicly
• A single machine can host multiple Virtual machines.
• A user using a virtual machine, will not realise if they are using a VPC or a real PC.
• VM Density, refers to - Number of Virtual machines running on a Host machie.
• This is an Unmanaged service

Steps to create a virtual machine:

• 1. Select the Subscription, Resource group and region
• 2. Select the image (OS + other SW)
• 3. Select the size
• 4. Create the admin user for the VPC
• 5. Expose any ports needed
• 6. Check the pricing and create.
• On creating a Virtual machine, all it's dependant resources will be created. Virtual machine, Network interfaces, Network Security groups, Public IP addresses, and Virtual netowrks...
• Then you can connect to the Virtual machine via, RDP (Windows), SSH (Linux), or Bastion.


• Automation: For automation of creating similar machines, use "Download a template for Automation" link to download the JSON file of the current configuration. It's called ARM (Azure Resource Manager) template. The template can be created from scratch, be exported, modified, uploaded or deployed. This can be source controlled.
• These are parameters.json and template.json. Template.json file defines the resources, and parameters provides, the configuration values.
• Create via SSH: az deployment group create --resource-group name-of-the-rg --template-file template.json --parameters parameters.json

Steps to delete a virtual machine:

Remember: A virtual machine when created, also creates many dependant resouces. So, to remove a Virtual machine, remove the resource group it is part of.

1. 1. Go to resource group page, and select "Delete resource group"

Cost of a virtual machine:

1. 1. Virtual machine type
2. 2. Disk allocation, and usage
3. 3. IP address
4. 4. Storage accounts cost

How to reduce the cost of a virtual machine:

1. 1. Auto shutdown (Available while creating the machine): Automatically shutdown machines when not needed (for Test and development machine)
2. 2. Reserved instances: Machines rented for 1-3 years and run continously (for production servers). Has great discounts and payments can be made monthly. Cannot be stopped / refunded.
3. 3. Spot instances: Machines that run on unused capacity from Azure. Can be evicted on any momemnt. Offers upto 90% discount. Good for batch processes, long running processes that can restart.
4. 4. Disk optimization: Choose the right storage disk. Default is premium SSD's for performance. Use standard SSD to reduce cost and for operations that are not processor intensive.
5. Tip 1: Select the right size machine
6. Tip 2: Select Linux machine over windows machine
7. Tip 3: Check prices in nearby regions

Availability in Azure:

1. 1. Fault domain: Group of physical hardware and a network switch. Similar to a RACK.
2. 2. Update domain: Logical group of physical hardware that can undergo maintenance and be rebooted at the same time. Done by Azure.
3. 3. Availability set (It's free): A collection of fault domains adn update domains. Can be upto 3 fault domains and 20 update domains. All domains are in the same zone (datacenter).
4. 4. Availability zone (It's free): Physically separate zone within an Azure region. Techinically, a different building containing autonomous data center. Provides protection against a complete zone shutdown. Deploy load balancer, to route betweem the VM's.

Virtual machine Scale Set (Free):

1. A group of separate VMs sharing the same image
2. Managed as a group
3. Can be scaled out or in manually or according to conditions specified
4. Great for handling unpredictable loads
5. Note: To make changes to the Scale set the original image has to be modified. Changes could be lost when it's made to indiviual machines, when they are scaled in.
6. For web apps, load balancer must be infront of the scale set.

To enable scale in / scale out feature, one must register under Subscriptions -> Resource providers -> Search for "microsoft.insights", select and click register.

Command prompt: az provider show --namespace microsoft.insights -o table

Steps to create a virtual machine scale set:

• 1. Choose a subscription, resource group, give a virtual machine name, region, availability zone or set, image, size, default user....
• 2. Scaling: Scaling can be done manually or by using custom autoscale rules. Select the created Scale set, and on the menu bar, choose "Scaling"
• 3. To add a custom rule, choose "Add a rule" from Rules form field.
• 4. Scale out / scale in can be done based on several metrics.
• 5. Scale in algorithms: Default (1. Instance Id that's greatest, 2. newest instance, 3. oldest vm...),

Azure instance metadata services:

1. 1. This is a REST API service
2. 2. Provides information about, SKU, storage, networking and scheduled events...
3. 3. Accessible only from the VM
4. 4. With scaleset, get notifications about upcoming evication, can be polled every minute.

How to use Azure instance metadata services?

• 1. Create a Virtual machine and select the VM and download the RDP file and connect...
• 2. Make a POST API request using postman, to the specific url, and Azure will respond with metadata about the current Virtual machine

Azure - App services

• 1. A fully managed web hosting for websites
• 2. Publish your code and it just runs
• 3. No access to the underlying servers
• 4. Secured and compliant, managed by Azure
• 5. Integrates with many source control and dev ops services like, Github, BitBucket, Azure DevOps, DockerHub and more...
• 6. Supported platforms, .NET, .NET Core, Node JS, Java, Python, PHP...
• 7. Also supports containers...
• 8. Supported application types: Web Apps, Web API's and web Jobs (batch processes)...
• 9. App service can be autoscaled to support spikes on loads
• 10. Extremely flexible, similar to VM Scale set...
• The main difference between App Service and the Virtual machine is that, when VM is stopped, the pricing also stops. But for App Services, whent he service is stopped, doesn't mean the billing stops!!!

Azure App Service - Create new:

1. 1. Search for, "App services", Click "Create new"
2. 2. Select subscription, resource group, give a name (unique across Azure), choose (publish code / docker container), Runtime stack (PHP, .NET Core..), OS, Region, Windows Plan, SKU (Stock keeping unit) and Size
3. 3. Click Review & Create

There's possiblitity to edit files deployed to App service via, a Console and an App Serice Editor window...

Scale up: Add more RAM, Processor and choose bigger and faster disks...

Scale out: Add more parallel machines...

Azure - AKS - Azure Kubernetes Services

• 1. Managed Kubernetes on Azure
• 2. Allows deploying containers and managing them using Kubernetes on Azure
• 3. Paying only on the instances (VMs) used

Containers:

1. Thin packaging model
2. Packages software, it's dependencies, and configuration files
3. Can be copied between machines
4. Uses the underlying OS

Containers Vs Virtual Machines:

• Containers are light weight as it's just a layer on top of the HOST OS and Virtual machines are heavy as it is a full OS
• Containers can be started and stopped in seconds to minutes, whereas starting Virtual machines takes a minimum couple of minutes for the boot to complete and the application to start.
• Containers can be restarted without restarting the hardware, but for Virtual machines some updates need harware restarts.

Problems with Containers:

• 1. Deployment: There could be multiple containers that might need deployment. And, this can't be done manually.
• 2. Scalability: Scalability must be done Automatically as well
• 3. Monitoring: How to monitor too many containers?
• 4. Routing: How do you route traffic across the containers?
• 5. Availability: How to make the containers run with high availability?

Deploying a container image to Azure:

• 1. Once the docker container is ready, we need to create a new Container registry
• 2. Choose a subscription, Give it a name, Choose a Tier (Basic, Standard, Premium), choose a resource group and location

To enable container deployment, one must register under Subscriptions -> Resource providers -> Search for "microsoft.ContainerRegistry", select and click register.

Kubernetes:

De facto standard for container management. Provides

• 1. Routing
• 2. Scalability
• 3. high availability
• 4. Automated deployment
• 5. configuration management and so on...

Command: az aks create -resource-group name-app-rg --name cart-aks --node-count 1 --generate-ssh-keys --attach-acr nameacr --node-vm-size Standard_DS1_v2

Azure - Azure functions

• 1. Small, foucused functions running as a result of an event.
• 2. Great for event driven systems
• 3. Automatically managed by Azure; It starts, stops and autoscales...
• 4. Flexible pricing plans
• 5. Serverless - A cloud resource that is completely managed by the cloud. The user need not think about VM's, CPU's, Memory etc...

Triggers and Bindings:

• Triggers: This makes a Function run. There are quite a few. Deeply integrated into Azure services.
• Triggers are not mandatory.
• Bindings: Are declarative connection to other resources. It can be input, output or both. These are not mandatory as well. Functions use bindings to read and write.

    namespace AzureExample.Function> {
        public static class EvenGridFunction {
            [FunctionName("EventGridFunction")]
            public static async Task Run(
                [HttpTrigger(AuthorizationLevel.Anonymous, "get, "post", Route = null)] HttpRequest req,
                [EventGrid(TopicEndpointUri = "MyEventGridtopicUriSetting", TopicKeySetting = "MyEventGridTopicKeySetting")]
                IAsyncCollector outputEvents, ILogger log) {
                    string name = req.Query(["name"]);

                    var myEvent = new EventGridEvent("message-id-3", "user-added", $"{{name: {name}}}", "event-type", DateTime.Utcnow, "1.0");
                    await outputEvents.AddAsync(myEvent);
                }
        }
    }
                            

Supported Trigger types:

• Blob storage
• Cosmos DB
• Dapr
• Event Grid
• Event Hubs
• HTTP Requests
• IOT Hub
• Kafka
• Queue Storage
• RabbitMQ
• Service Bus
• Timer

Supported Binding types:

• Blob storage
• Cosmos DB
• Dapr
• Event Grid
• Event Hubs
• HTTP Requests
• IOT Hub
• Kafka
• Mobile apps
• Notification Hub
• Queue Storage
• RabbitMQ
• SendGrid
• Service Bus
• SignalR
• Table storage

Azure functions: Cold start

Since Azure Functions is a serverless application, incase if the server is shutdown by Azure on low demand scenarios, it might take few seconds for Azure to bring the application back up.

This problem can be avoid using the right hosting plan. The plans are:

• Consumption: Pay for function running time, and network data consumed. Be aware of ColdStarts and Function cannot consume more than 1.5 GB RAM.
• Premium plan: Pay for pre-warmed instances. Pay for CPU duration and memory usage. You can choose the machine configuration options. Advantages: No cold starts, no memory limits, better application performance and VNet configurations available, the prices are predictable. Downsides: EXPENSIVE.
• Dedicated plan: Functions run on existing app service. Great if the app server is underutilized. Make sure "Always On" option is activated. There's no auto-scale support for Dedicated plan functions.

Durable functions:

• Stateful functions that react with external resources and keep track of flow
• Offer simple syntax, hide complexities of managing states and allows retries...
• Used usually with, function chaining...

• A network where one can deploy cloud resources
• Many resources (VM, App Services, DBs etc...) can be deployed within Vnets
• "Virtual" here means we don't create a physical network, but use existing Azure networks to create a logical network only for our application
• Resources in the same VNet can communicate with each other by default, but cannot communicate to resources in another VNet
• VNets are free, upto 50 VNets.
• VNets are scoped to single subscription and a single region
• Can be connected via Peering
• Segmented via Subnets
• Protected via NSG (Network Security Group) - Defines who can connect in and out of a subnet, like a mini firewall. NSG is free.
• NSG looks at 5 things called a Security Rule : Source, Source port, Destination, Destination port, Protocol
• Each VNet has IP range, upto 65,536, can be customized, and all devices must be within this address range
• Must be expressed in CIDR (Classless Inter-domain Routing) notation

Network Peering
Securing the VM Access
Service endpoint:
Private link:
Service endpoint / Private link VS VNet Integration:
App Service Access Restrictions:

How to configure a Application gateway

• Backend pools:Can be VMs, App services, or scale sets...
• HTTP Settings:Settings for the incoming HTTP requests
• Frontend IP Configuraitons:Public IP exposed by the Application gateway
• Listeners:Receives requests on a specific port and protocol
• Rules:A rule connecting the listener to the backend pool

Application Gateway and AKS: Application gateway doesn't support AKS as of now.

Application gateway and Functions: Function apps are basically app services. App Gateway can protect function apps.

Azure provides many data solutions as cloud services that are fully managed, and can be part of Azure app or fully independant. Some of them are:

• Relational databases
• NoSQL databases
• Object stores

Major database features:

• Security: data should be encrypted on rest
• Backup: If something fails, we should have backup
• Availability: The service should have high availability

Tiers of Azure Blob Storage