Header Ads

Compute - Microsoft Azure

Azure Compute provides infrastructure for your new or existing applications. Based on your requirement you can decide what type of infrastructure you need. It will also help you scale your application during peak hours or during high demand by adding more instances (scale-out) or by upgrading the plan (scale-up).

Before we can dig down all the services, it important to know a few technical terms such as:
  • Scale-out and Scale-in: Scale-out is the process of adding more instances of the machine or web app instances within the app service plan, and adding a load balancer to distribute the demand. This is also known as vertical scaling.
  • Scale-up and Scale-down: Scale-up is the process of upgrading the service plan or machine's hardware such as CPU or RAM, to accommodate more traffic. Scale-down is exactly the opposite. This is also known as horizontal scaling.
  • Autoscaling: It is a process of scaling (scale-out/scale-up) your application automatically during high demand by adding some rules/triggers. It can be done either from the portal or by using scripting tools such as PowerShell or Azure CLI.
  • Fault domains: It is a set of hardware components such as racks, storage, network switches, and a server, that share a single point of failure. Microsoft doesn't have any control on this as it is unplanned or unpredictive outrage.
  • Update domains: It is one logical unit of deployment. Microsoft has full control over update domains and it ensures that update to the hardware components will have the least impact on your application.
Few services under compute category are:
  • Virtual Machine: It helps you create and deploy, highly configurable Windows or Linux machine.
  • Virtual Machine Scale Sets: It allows you to autoscale your application during high workload by deploying up to thousands of VMs. You can select the minimum and maximum number of VMs to be deployed. Autoscaling can be done by configuring the rules such as CPU threshold, memory usage, data in etc. You can also select load balancing option such as Load Balancer or Application Gateway.
  • Function Apps: It is event-driven, serverless computing which gives your application an extra edge. The event can be triggered from HTTP request, time-based (Cron), when a message is added into queue or service bus (topic and queue), etc. It is kind of nano-services, where each function acts like an individual service. You can use this for cleaning up records, analysing requests, configuring apps, etc.
  • App Services: It helps you in building, deploying, and scaling your web, mobile, or API application. This is part of PaaS offering from Microsoft, it gives you high performance, scalability, security without even worrying much about backend infrastructure.
  • Container instances: It helps you in creating, deploying applications on containers with Windows or Linux images, without provisioning VMs. It also makes sure that your container group is isolated from other container groups.
  • Batch: It can run large-scale parallel compute solutions and HPC (high-performance computing) applications. You can also configure to run commands or scripts or jobs on a given time such as scaling your applications on weekends to handle a huge workload or running thousands of machines parallelly to execute HPC workflow.
  • Azure Kubernetes Services (AKS): You can orchestrate hundreds of containers with the help of AKS. AKS helps you in scaling, managing, discovering, securing, and communicating with other containers.
  • Service Fabric: It enables you to build, package, deploy, and manage scalable and reliable microservices. It was initially built for microservice kind of architecture, where it provides lightweight runtime to build distributed scalable, stateless or stateful microservices.
  • Availability Sets: It is a group of VMs deployed across fault domains and update domains to make sure that application is available all the time. It also ensures high availability of your application by removing single points of failure such as network failure, or power outages.

No comments

Powered by Blogger.