Cost Optimisation in Azure

-


As we go through the current situation with coronavirus many companies will need to review their spending.  I have gathered some key resources and tips to start making real differences to your Azure costs today!

This is by no means an exhaustive list and I will add to this as I remember stuff over the coming days but I wanted to share this now so you can start to help your business save money as soon as possible.

Monitoring

This may seem an odd way to start a cost saving exercise but the key to knowing what actions to take is having information available to you so you can make informed decisions.  These metrics can give you great insights into your environments and applications and help you identify underutilised resources.

All resources come with some key metrics available to you out of the box without enabling or setting up other resources. Simply browse to the resource in the portal and you will see the metrics on the overview page or can go into the metrics section of the resources.

 Azure monitor covers a range of tools to collect telemetry from your resources.

Log Analytics can be used to collect a wide range of telemetry from resources and allow you to create custom queries and get greater insight into your resource usage.

Application Insights provides insight into your application usage and can help you identify trends and troubleshoot issues quickly.

Autoscale

Azure Monitor Autoscale applies only to Virtual Machine Scale Sets, Cloud Services, App Service - Web Apps, and API Management services.  If you are using any of these services consider using the Autoscale feature to monitor the performance of your apps and scale up or down.  This can scale to meet additional load but equally can be used to scale down your resources to save money when your applications are not as busy.

You can scale on a schedule, so maybe overnight when your resources are less busy.  You can also scale depending on metrics for example, you could scale down if CPU is below 30% for ten minutes, you then set the number of instances you wish to scale by when this threshold is hit.  You can set your minimum and maximum number of instances. This is a great way to automate the scaling of resources to help save money.

Monitor your spending

You can use Azure Cost Management to keep an eye on your spending in Azure and AWS.  You can filter your spend by subscription, resource group, resource type etc.  This is where tagging resources is a big benefit. If you add tags to resources for example for cost-centre, projects, teams etc you can be more granular with your filters to view your spending by these tags, this will help identify trends and identify areas where you may be able to save money.

Right size and remove unused resources

Under Cost Management in Azure you will find Advisor Recommendations.  This will help identify idle and underutilised resources allowing you to remove or scale down resources to save money.




The recommendations also include other information to help you make cost savings like using reserved instances I provide a real example of the savings that can be made in the section on predict your spending.

Set Budgets

In cost management you can set budgets for subscription or resources groups and receive alerts if these trend in a way that would exceed your budget for a given period.  This allows you to react to this and take action to get spending back on track.

Automate the stop/start of Virtual Machines

Using an Azure Automation account you can create your own scripts or use runbooks from the market place that can stop/start your VMs.  For example, you can use this to ensure dev/test machines are only online Monday to Friday.  When VMs are in a deallocated state you do not pay for the compute resources, only for the disks. Turning off machines when you do not need them can help you realise significant savings. 

https://docs.microsoft.com/en-us/azure/automation/automation-solution-vm-management

Predict your spending

Before you deploy new solutions/resources make use of the Azure Cost Calculator  to get a good understanding of what the cost of running these resources will be.  The calculator provides options to see the cost saving if reserved instances are used, also if you already have license agreements from Microsoft make use of the hybrid agreements.

As an example of saving that can be made on a Virtual Machine (price correct at time of writing);

 Virtual Machine SKU D2 v3 2vCPU 8 GB RAM - UK South
1 x VM running for 730 hours would cost
£113.7 per month

Using a 3 year reserved instance saves 32%
£77.16 per month

Using hyrbid Licensing and 3 year reserved instance saves over 75%
£27.10 per month

Restrict the use of expensive SKUs

To control spending you can use Azure Policy to restrict the deployment of certain SKUs on resources to ensure staff can not deploy expensive SKUs that could allow spending to get out of control.

Use the right resource types

Review whether you can you save money by changing the resource type you are using to deliver your services.  For example, can a webapp be used instead of a VM for hosting your site.  Can you use Azure Functions instead of running these scripts on VMs.

Enterprise Customer Saving

Enterprise Agreement customers can realise saving by running dev/test resources in an Azure Enterprise Dev/Test subscription, for full details of the benefits click here.

If you take action on some of these points you can start to realise cost savings now.

I hope this helps!

Bob


Comments

Post a Comment

Popular posts from this blog

Terraform Functions - Part 3 - Conditional expression

-
In part three of this blog series we will take a look at the conditional expression function in  Terraform  and see how we can use this with a basic  Azure Web App  deployment. In a future blog we will explore how to use the conditional expression alongside the count function in a more complex resource deployment.  In part one we cover an overview of the blog series and some pre-requisites you will need to get started when working through the deployment steps in this blog series, if you haven't seen this information please review this  here  before you get started with the steps below. Conditional expressions can be used to dynamically omit or change an argument depending on a certain condition. This can be useful for several scenarios, one example may be that you create a module for deploying VMs and have a variable for VM size of small or large, you can set this as a variable and evaluate this variable to change configurations like SKU, disk size etc.  The example we will go thr
Read more

Azure Kubernetes Service (AKS) and Managed Identities

-
Image
In this blog I will be exploring the use of Azure Manged Identities in Azure Kubernetes Service (AKS).  We will then discuss how we can use managed identities according to security best practice. We will look at how we configure the managed identities for the AKS cluster so it can in turn manage other Azure resources.  We will explore how we can configure managed identities for our services/applications that are running on AKS so pods can reach out to other Azure services. Managed Identities Why are we using managed identities? The alternative is to use Service Principal accounts (SPNs). The issues with SPNs is you have a client secret which you have to manage and keep secure. Your cluster apps and services will need to access the SPNs you have created so this means potentially saving it in a few places so it is available to CI/CD pipelines. The secret attached to an SPN rotates so you need to ensure it is valid to ensure your cluster and services continue to run. Managed identi
Read more

Working with WSL and AKS

-
Image
Introduction I find I am working with  Azure Kubernetes Service (AKS) more and more recently and I thought I would share a few tips and snippets of code I have found useful. For reasons I will explain shortly I have started using the Windows Subsystem for Linux (WSL) as my main way of managing AKS.  If you are starting out with AKS I hope you will find some of these hints and tips useful. AKS is a fully managed Kubernetes service from Azure, if you want to find out more about AKS check out the material and video from Microsoft here  .  For the purpose of this document I am assuming you are familiar with AKS and have at least started to play around with it.   Commands There are some key commands you will need when working with AKS and kubectl you can find some of these on the first link below.  While working with AKS you will  be using other tools like Docker for creating and managing your container images.,  I have provided some key starter commands for this on the second link. Helm
Read more