Can you autoscale on demand without worrying about infra expenses. If not, chances are you are worried about overspilling your operational budget and that means you lack a system that manages your cloud expenses effectively.
This system is called Financial Operations on cloud in short FinOps. Here are 5 effective FinOps frameworks that will help you to make the most out of your infrastructure budget.
Why FinOps ?
Every business in the current digitally transformed world wants to become bestsellers of quality products or services. However, some frequent events, if overlooked, can impact this ambitious journey. These are
Overspilling infrastructure expenses
Underutilised monolithic systems
Infrastructure bottlenecks for business scale and performance.
Any business, no matter how big or small they are, always run on some set operational budget. Nobody has an unlimited credit card. To make the most out of their infrastructure budget, they need to
Build financial visibility.
Optimise resource utilisation.
Formalise accountability for operational expenses on Cloud.
Doing effective FinOps can be overwhelming, especially when you are spoiled for so many options on cloud. However, it's not an impossible task, just needs to follow certain structure and due diligence in planning.
Here are some of the key factors that broadly drives FinOps for Modern Infrastructure and can heavily impact infrastructure budget if not handled properly.
The Key Factors That Impact FinOps in Modern Infrastructure
As you can see each of the 4 factors above have more than one configuration options. You need to choose carefully, so that you can don't end up overspending or underutilising the budget required for scalable modern infrastructure. Here's what should be followed
Billing Method
To avoid overspending or underutilised infrastructure budget choose a billing method that will save money in the long run. Most cloud providers offer two billing options as below:
Postpaid/Pay As You Go | Prepaid |
|---|---|
It means use first and then pay by the no. of hrs/secs used resources. This the default mode of billing in all cloud providers available today. If you don't specify any billing method it uses this option by default. | This is a commitment or subscription model of payment. Let's say you want to run some services on cloud , so you opt for a monthly or yearly subscription and get discounts on the overall price. The rates of discounts varies on the duration of subscription or commitment, the longer the timeline the better the discounted rates. Sometimes it can be as much as (30 - 50)% of the base price. |
Recommendation: Prepaid model.
Even if it's for a month, you will see significant savings in your expenses with the discounts that come in even for a month. So it's better to set the default billing method to prepaid or subscription while provisioning infrastructure.
And let's say you tear it down in a few days, there are refund policies for unused credits in most cloud providers. So there is no harm in going with a default prepaid billing method.
Optimisation
Even though prepaid billing method is effective, however it alone cannot help impact spending unless infrastructure utilisation is optimised as well.
Infrastructure should scale when demand rises and scale down if it's is low. Otherwise, you will end up paying for underutilised resources which can become pretty expensive, sometimes even crossing your revenue margin.
Scaling infrastructure is the easy task, what is tricky is scaling down underutilised resources and this is determined by the method used for infra scale
Vertical Scaling (Scale Up) | Horizontal Scaling (Scale Out) |
|---|---|
Scaling by ramping up the sizes of existing systems by adding more resources such as cpu, memory etc is vertical scaling or scaling up. With this method you can easily allocate more power to your infrastructure when needed. However, once allocated these resources cannot be released or reused without decommissioning the systems.Till then it will be reserved for the system it's allocated to. | Scaling by adding more systems in a cluster is horizontal scaling or scaling out . Let's say you have a Kubernetes cluster which is reaching 90% cpu/ram utilisation in peak hours, so you scale out by adding more no. of nodes. So what you can do is set a minimum and maximum no. of nodes and the cluster will scale out and down with when it reaches threshold limits. This is called horizontal scaling or scaling out. |
Recommendation: Scale out instead of scaling Up.
Use automatic horizontal scaling so you can scale out and scale down based on performance thresholds especially in cloud managed compute cluster services.
This will help optimise resource utilisation in modern infrastructure and avoid scenarios of overspending or underutilisation.
Service Type
Services in modern infrastructure can be setup and managed in two ways -
Do It Yourself or Let a Service Provider do it for you. Both has its own share of advantages and disadvantages be it in terms of compliance and security , operations or FinOps. Let's see what these impacts are
Service Catagories | Pros | Cons |
|---|---|---|
Managed Services |
|
|
Self Managed Services |
|
|
Recommendation: Managed Services with enterprise subscription from security & compliance ready providers.
Although prices might seem higher, with prepaid billing it becomes same as self managed systems.
As you can see, both service categories have their own share of pros and cons, so you have to select which is a priority for your business.
Some businesses prefer to choose self managed model for data localisation, privacy and security regulations. They are basically worried about IP security on systems managed by someone else.
In a such case, you can actually discuss with the cloud provider and get them clarify their compliance readiness in an agreement. With a little due diligence managed services are recommended for higher benefits like High Availability, Autoscaling, Support & Maintenance.
Budgeting
This brings me to the last and the most important factor in FinOps - Budgeting, one of the major requirement that every organisation faces every financial year. And if you are a cloud user, you must have used this tool at some point of time in your cloud journey.
It's the Cost or Pricing Calculator, a tool provided by many cloud providers for calculating cost of infrastructure resources they provide. For instance, you want to know how much will a large compute instance with 60 GB storage will cost, the cloud cost calculators will do that. If you have not used one before here are some links for reference:
Cost or Pricing calculators are great and undoubtedly powerful tools when it comes to budget estimation in FinOps.
However, what I find overwhelming is the wide range of pricing options. AWS EC2 service itself has 14 instance family, each family comprises around 10 instance types, so if I have to calculate price of a compute server I will have to choose from 140 options. It's like a kid in a candy store, spoilt for options.
To address this challenge, I prefer creating a Rate Card which will contain some selected instance types that will be used a the base infra types.
For instance, let's say you have a memory intensive application, so the rate card will offer only memory optimised instance family with T - shirt sized instance types. Same
can include network and other services as well. One of my sample rate card looks something like this.
Such rate cards can be easily used by FinOps teams, even if they are not technical.
Once you create rate cards budget estimation becomes very efficient and accurate. In-fact you can then have a continuous review of your expense estimation based on exact requirement instead of speculation. This will avoid over or under estimation in FinOps for modern infrastructure.
One common myth about FinOps is that it's a management task, something that can be only done manually in documents. That's not true, some of the factors discussed above can be automated as a part of infrastructure creation so that these practices become default configurations during infrastructure provisioning. Here's how you do it.
FinOps As Code:
Infrastructure-As-Code to:
Set default subscription for infrastructure provisioners.
Use billing apis to convert pay-as-go services to subscription.
Pipeline As Code to bulk deploy of billing methods changes.
Schedule jobs to stop or terminate unused infra.
Here's my recipe of Effective FinOps for modern infrastructure
Always review your usage and rework the budget.
Utilize support points.
Inform, optimise and operate.
Don’t be alarmed with initial expenses as most subscriptions deduct money few days before the month ends. You can get refunds from the time you stopped the service.
Hope you find this blog useful. All the best for your next budgeting exercise.
If you like this article, I am sure you will find 10-Factor Infrastructure even more useful. It compiles all these tried and tested methodologies, design patterns & best practices into a complete framework for building secure, scalable and resilient modern infrastructure.
Don’t let your best-selling product suffer due to an unstable, vulnerable & mutable infrastructure.
Regards
Kamalika Majumder
Comments