Microsoft Azure Optimisation tools we utilise

We have seen savings of up to 70% in an initial optimisation. Of course, your results may differ, but suffice to say that almost every first optimisation has led to substantial savings. By continuously optimising, improving visibility and establishing governance, you can maintain an optimal state so costs do not get out of hand in future.

It is also important to note that it’s not always about savings. While cost takeout is a sensible place to start, it’s important to think about value. When you start
thinking about cost to serve, such as the cost per new client or cost per sale, you get a better understanding of how to use the cloud more effectively. You can
make more informed decisions on where to focus your cloud spending.

Here is one example, if consumption costs double for an application, and sales triple, you have derived more value from your cloud spend on that application and
achieved a better cost to serve.

To achieve initial savings:

• Analyse your estate and identify optimisation opportunities across compute, storage and networking focusing on the biggest costs first and/or what you see increasing month over month. You will need access to the costs and utilisation of the Azure services. If you have tagged the Azure services by environment for instance the optimisation will be easier. Also consider strategies around discounting mechanisms such as reserved instances.

To maintain an optimal state, you will need:
• Governance – the policies around budget adherence, resource creation permissions and more
• Transparency – reports of costs, usage and forecasts
• KPIs well defined – understand and work towards specific goals
• Resource tagging – identify usage, ownership and department allocation if you use a chargeback model
• Continuous review – keeping on top of optimisation opportunities and new cloud features is key. As the cloud evolves, your portfolio must evolve alongside.

You need to start optimising as you move to the cloud. Optimisation must be continuous and part of your operating model. It is not a one and done event of once optimised, always optimised. By applying optimisation practices as you move into the cloud you establish a culture of optimisation and accountability. You need to setup the org structures, define the optimisation roles, rightsize on entry, and define policies and processes as you move to the cloud.

If you wait until you are in the cloud and with a significant number of assets in the cloud it could take months and during that time you could lose money if you drift from an original optimal state. We do see clients who start optimising once in the cloud. This typically happens often after costs get out of control. In these instances, it’s recommended to identify and implement the biggest savings then put in place the roles, governance, reports and process to stay optimal.

There is no particularly defined cycle. How often you optimise depends on your needs such as the velocity of your cloud adoption, the speed at which you develop and the alignment of your financial cycles.

We typically audit monthly but the tools we use allow for real time analysis. In some instances, we will look on a daily basis for certain spikes in usage so they can be addressed quickly. It’s important to stay on top of everything during the initial phases of optimisation in order to keep budgets under control. Automation can help you here. As you move to the visibility and governance phases, this becomes continuous monitoring, so spend is constantly monitored and spikes are instantly identified.

When you originally moved to the cloud and rightsized on entry, you were likely optimal… at that moment in time. However, as needs change, application usage changes and new resources are added, the need for continuous optimisation grows. The cloud provides a lot of freedom but requires a balance across the often-competing priorities of cost, performance and speed. Optimisation and governance enables you to maintain an optimal state.

Gaining visibility from Azure Cost Management (ACM) and recommendations from Azure Advisor is only half the battle. In fact, it’s less than half. ACM does not have all the reports we know our clients need. Particularly valuable are our month over month reports which show changes over time, and our reserved instance (RI) analysis that provides an understanding into efficient usage of RI and insights into where it could be used.

In addition, the advice from Azure Advisor must be complemented with an understanding of how your apps and data are used in order to understand if those recommendations are correct. Our extensive experience gives us an informed approach with 50 scenarios to evaluate for optimisation including VM usage, storage usage, orphaned resource handling and many more.

We will also build your tagging strategy, cost allocation strategy, policies and more aligned with your business and technical goals so you can maintain an optimal state

Standard Azure tools such as Azure Cost Management, Azure Advisor and Azure Monitor are a great start and will pick up some of the quick wins. But they are not complete. In order to cover all types of optimisation you need a good knowledge of the Azure Advisor algorithms to understand if/how you should leverage the recommendation.

For instance, when recommending VM rightsizing, Azure Advisor only looks at the last 7 days of usage and CPU utilization. You may want to base your recommendation on a longer time period and also look at memory utilisation, disk utilisation, network utilisation and understand if there are any attached disks that are in use.

Common resources may be distributed or buried in central IT costs. If you share the costs today, you can bring that same model to the cloud. If you don’t share costs today, we can create that model in the cloud for you. So, if for instance, the shared resource(s) is an application that tracks the calling systems (e.g. authorisation system used by multiple applications) then the costs can be allocated proportionately to the calling systems. A well thought out tagging strategy that is built around your business gives you the visibility you need to understand where shared resources are being consumed, and accurately allocate costs.

This will depend how you want to view your data, what you need for your KPIs and how you want to manage accountability.
For instance, some suggested tags at a minimum are:
• Environment – such as dev, test, production
• Category – such as infrastructure, application
• Application – the application name

You can always join these tags to an external database keyed by application for instance. This will allow you to:
• Understand the associated department for chargeback purposes and budget limits
• Associate with a business process and the number of customers or sales from that business process
• Derive business KPIs such as consumption costs per unit of value

This tagging approach will shed light on the real value you are getting from Azure, allowing you to identify where expectations are being met, and where they are not.

Some examples are:
• Storage account blob tiering – use the hot, cool and archive tiers appropriately
• Right size SQL databases
• Reserved instances for SQL databases
• Hybrid benefits for SQL databases
• Some regions have a lower cost
• Go serverless where you can

There is some automation within Azure Advisor today. We bring additional automation tools and capabilities over and above what’s available within Azure Advisor and Azure Cost Management. We can create specific dashboards with specific alerts that incorporate a combination of factors – usage, cost, time, tags, etc.

A tool will not know how the business uses an application. In addition, you may not be familiar with the algorithms used in the tool and therefore may not be confident in a recommendation. Here are some considerations and differences between primarily a tools-based approach to optimisation such as relying primarily on first or third party tools and a more comprehensive services approach from SR Cloud.

Tool based approach

• A tool will help you to find the low hanging fruit and act fast to make some quick wins.
• You can use a tool repeatedly but bear in mind its value will diminish quickly with every subsequent use.
• Be wary about becoming over reliant upon a tool. You may not understand how it arrived at any recommendations (e.g. did algorithm look at past 7 days or 30 days, can you adjust it, does it look at CPU/ Memory/Disk/Networking for resizing or just CPU) and have full confidence in any recommendation.
• A tool will analyse usage for you but can’t uncover real insight.
• You still have to create the optimisation strategy and ensure role-based access control and tags are in place.

Service based approach from SR Cloud

• A tool is one part of the solution but not the sole focus. You are left with the tool if you do not retain SR Cloud to continuously optimise.
• We help you define the best optimisation strategy and approach for your business.
• We help you define governance policies and guidelines to get to and maintain an optimal state.
• We understand the logic in the tools used to create recommendations.
• We work with technical and business teams to create and approve recommendations that are aligned with client goals, application performance, security, availability and more.
• SR Cloud can also act as the project management office to manage the optimisation cycle.
• SR Cloud can implement the right recommendations.
• SR Cloud can evaluate the portfolio for further optimisation such as going serverless, using containers, modernising apps to use PaaS instead of an IaaS approach

A workshop is our recommended way to start to understand the situation and provide quick win recommendations. It helps you understand what you have to gain.
The workshop is a three weeks collaborative effort with:
• Week 1 – being the analysis phase to review cloud costs and usage and design the first iteration of a dashboard
• Week 2 – identifying quick wins; deploying a first version of a dashboard; defining strategies for tagging, allocation and more
• Week 3 – implementing quick wins where we can and build the roadmap for short and long term optimisation

And, we can tailor this to any specific needs to include a strategy for security and regulatory compliance for example.