Year after year, a large number of engineering teams around the world dedicate their time and energy to cloud optimization, still the research suggests that most businesses are not able to keep up with the surging cloud costs. What exactly is going wrong for these businesses? Did you ever wonder what might be the solution?
There are a number of reasons why this is happening and will continue to happen if we do not collectively work on a solution. Low-cost visibility, tricky cloud pricing, and problems figuring out future demand are some of the reasons. As a result, businesses waste cloud resources, time, and money due to these impending issues.
It is challenging to interpret a cloud bill and cloud service providers are unable to help in cost reduction. Read more to find out the most thriving issues that waste cloud resources, time, and financial resources, and find solutions to them. Conneqtion Group is India’s 1st Oracle PaaS Partners and we are dedicated to help you reduce your business costs by shifting your business to the cloud, in a sustainable way.
In this blog, we will cover:
- Why is cloud cost optimization so overwhelming?
- The 6 Top Cloud Cost optimization problems to fix in 2022
Why is cloud cost optimization so challenging?
The pay-per-use model of the public cloud has imparted a sense of freedom to engineering teams, but the freedom is proving to be expensive. It is a challenge and check out cloud cost reduction stories from Pinterest to get a better understanding of the subject.
Research suggests that most businesses find it difficult to handle the rising cloud costs. A conventional public cloud spend surpassed budget by an average of 24%. There are various reasons why it is still happening. These often come down to low cost visibility, complex cloud pricing, and issues with forecasting future demand. All these problems might be contributing to how much your business is spending and will need to spend in the future.
Cloud providers have understood this and they are simply not helping either. Decoding a cloud bill can be overwhelming for most businesses. It can be so daunting that some engineering teams choose to ignore the cloud bill which comes back to bite them at the end of the month.
If you want to reduce your cloud costs, read on.
The 6 top cloud cost optimization problems to fix in 2022
#1: Still getting attracted by reservations and savings plans
The first solution when thinking about reducing the cloud costs is to pay less for the services your team currently uses. Businesses select savings plans or reservations as they come with notable discounts when compared to the on-demand pricing model.
It is amazing to pay upfront for a predictable cloud bill.
Though if you notice, you will find out that instead of solving the problem, you are simply getting a discount on the problem and also committing to it for the next few years. Your business doesn’t deserve that.
What is the issue here?
If you read Pinterest’s story of Cloud cost reduction, Pinterest had paid upfront $170 million for AWS services in advance but still had to shell out an extra $20 million for additional resources. That’s more than 10 percent. Now, Pinterest may not mind but your bootstrapped startup can be toppled by these unprecedented costs.
The point is, Pinterest’s story allows us to understand that knowing how much capacity you will require in 1 to 3 years is difficult.
By committing to a single provider for a long time, there is no flexibility, your business is locked in, and you may need to shell out a hefty price for evolving business requirements.
How to tackle the savings plans?
You guessed it right, the best solution is to avoid savings plans if possible.
Try not to purchase resources in advance, at the same time think about different options that solve the cloud spend including:
- Resource scheduling
- Bin packing
#2: Succumbing to the trap of overprovisioning
Overprovisioning happens when your business selects resources larger than required to run the workloads. There is a safety mindset at work here, as no team wants disruptions in the middle of a critical project.
In some organizations, teams are habituated to have more resources than required as a “backup”. Though performance-wise this option is perfect for engineers, it takes a toll on the operational costs of an organization by increasing cloud waste.
What is the issue with overprovisioning?
The easiest answer is that overprovisioning leads to cloud wastage and needless expenses that can quickly go to the roof.
Encouraging overprovisioning in your teams can be a terrible idea in the future. If the team gets used to getting an instance bigger than the workload requirement, think how this works when you scale your business. The cloud cost will take a huge chunk of your business revenue.
Don’t you think it would be better if you could spend this money on something substantial that helps you in the long run? For example, on branding and marketing so that you can establish your business as a credible source in your industry.
How to handle overprovisioning?
The typical approach was to spend money on custom monitoring and cost management solutions. By rightsizing recommendations, your business can help decrease the team’s dependency on overprovisioned resources. However, you will still need to implement them manually. Another solution would be automation. Automated rightsizing can select the most probable instance types and sizes that meet your application’s requirements while keeping costs under control.
When a cluster requires additional nodes, the AI-driven instance selection algorithm automatically chooses the resources that drive ultimate performance. And do you know the best part of this? Your team doesn’t have to spend time and energy to sort this as it is automated.
As overprovisioning is nested in the business culture, it is critical to create cost optimization routines to enhance performance, reliability, and reduce costs. Automation can be the first important step to organizational change.
#3: Getting affected by unused cloud resources
It is easy to create an instance for a project and then forget to shut it down.
As a result, most teams struggle with unused resources that have no work but still continue to add to the cost. Check out Adobe’s mistake, which cost them a whopping $80,000 a day in unplanned Azure expenses. These are the types of mistakes that your business cannot afford.
This problem is particularly specific to large businesses with various events taking place at the same time, with a lack of resource visibility.
Initiatives managed outside the realm and knowledge of the IT teams can account for 40% of all IT expenses at an organization. Moreover, statistics show that shadow cloud usages can be upto 10 times the size of known cloud usage.
What’s wrong with unused cloud resources?
Unused and unmanaged cloud resources contribute to costs with complex sustainability outcomes.
Having said that, data centers use a lot of electricity and hardware, impacting majorly to the carbon footprint of the business. The energy requirement doubles roughly every four years, and every new operation from AWS of Azure contributes to this problem.
Hence, reducing cloud waste is important to reduce unnecessary spending and related carbon footprint.
How to tackle this issue?
By making sure that you only use the resources you truly require may be overwhelming at first, especially in large corporations.
Teams working on e-commerce infrastructure would have an idea of how fast things can change. For instance, a new influencer mention can bring in hundreds and thousands of new sales or a website/app that goes down from the rise in traffic.
#4: Handling drops and spikes in demand inefficiently
Most other apps experience these changes in usage over time, but maintaining a balance between performance and costs is still a struggle for most companies.
What is the issue?
Rising traffic can generate a huge cloud bill if you don’t keep an eye or cause your app to crash if there are strict limits on the resources. So, when there is low demand, there is always a risk of overpaying. On the other hand, when the demand is high, the service quality may be impacted.
Agreed, there are cloud cost management solutions that can monitor your usage and notify you in real-time if it goes beyond permissible limits. These tools can offer terrific recommendations on how to adjust your cloud resources to the changing demands. Though, manual cloud management is overwhelming and time consuming.
These are the things that you need to know:
- Carefully handling traffic surges and drops – respectively scaling resources up/down across all the services.
- Making sure that changes applied on one workload doesn’t impact other workloads.
- Optimizing and managing resource groups manually to know that there are resources ideal for your workloads.
What is the solution?
This is another area where cloud automation can be used to help you save time and resources.
By autoscaling, your business can manage all the tasks mentioned above and keep cloud costs under control. If you leverage container orchestrator Kubernetes, you get three advantages:
- Horizontal Pod Autoscaler (HPA) adds or gets rid of pod duplicates to match your changing usage. It studies your application to identify if the number of replicas should change.
- Vertical Pod Autoscaler (VPA) reduces/increases CPU and memory usage requests to help you align your resources with current usage.
- Cluster Autoscaler changes the number of nodes in a cluster on supported platforms.
#5: Denying the leverage of spot instances
Cloud service providers offer their unused capacity at massive discounts, if you compare it with their on-demand pricing. In AWS, spot instances are available at 10% of the regular price.
What’s challenging about spot instances?
As you bid on spare computing resources, there is no way to know how long the capacities will be available. Spot instances come with a default duration; for instance, AWS provides a spot instance that offers an uninterrupted time guarantee of more than 5 hours.
Apart from that, service providers can reclaim the spot instance and offer you a short notice of only 2.5 minutes.
It is not feasible to make changes in that period of time. Developing a new VM also takes more time so there is a risk of potential downtime.
Hence, if you wish to use spot instances, it is best to know that interruptions will occur. They are not the ideal candidates for critical workloads.
How to manage this issue?
Despite the above mentioned issues, spot instances are amazing for services that can have more than one replica. It is good that most services are stateless in contemporary architectures, because Kubernetes was built for this kind of setup.
Here’s the whole process of working with spot instances:
- It is important to qualify your workloads and how well it manages interruptions.
- Then you should look for instances that your vendor provides. Try to select the less popular instances and monitor their interruption frequency.
- Then it is time to set a max bid to avoid interruption if the price instantly shoots up.
- It is best to manage spot instances in groups and request multiple instances to increase your chances of getting them filled.
You can perform these steps manually, but to make it work, be ready for a large configuration, setup, and maintenance tasks.
#6: Postponing Cloud Automation
Automation helps in these cases. Apart from the process mentioned above, automated solutions can assist instantly when spot instances are not available.
If you have handled cloud-native technologies, automation is probably the best option for your business. Cloud automation offers amazing benefits for IT teams in large enterprises.
Firstly, it reduces manual labor of configuring virtual machines, creating clusters, or selecting the right resources. This change helps you save time and allows your team to handle important tasks and make the best use of your cloud infrastructure.
Moreover, automation tools provide frequent updates which are crucial for deployment. It also significantly decreases the odds of human errors and reduces infrastructure costs, and improves backup processes.
Lastly, automation allows you to gain visibility of resources in use throughout your business that would be difficult to manage. In a nutshell, cloud automation is the new normal in the technology industry.
Why should you not delay cloud automation?
If cloud automation offers so many benefits, why are not all businesses following them?
Automation can be overwhelming, from the resistance to new solutions to fears that may cost a lot to implement and the need to update the current setup.
Similar to other digital transformation projects, the main point is to encourage change on a human level. When it comes to accepting automation, a research from Mckinsey proves that workers fear that they can be replaced by technology and hence this issue needs to be handled with care.
At the same time, automation brings in various benefits. But by delaying cloud automation in 2022 means your business can miss out on:
- Picking the best instance types and sizes for your workloads.
- Autoscaling cloud resources to manage rise/drops in demand.
- Getting rid of resources that are not being optimized to reduce costs.
- Improving spot instances by handling potential interruptions.
- Decreasing unnecessary expenses in storage, backups, security etc.
What is the solution?
The problem of handling the workforce against automation is an old problem and businesses in the past have suffered due to carelessness. It is best to remind your teams that cloud automation will liberate them from boring, repetitive tasks and they can channel their energy on innovation.
So, how should you find and retire unused instances? This is where our new friend Automation comes to the rescue again.
Automated cloud optimization solutions can monitor your cloud usage against inefficiencies and compact resources. They can automatically retire unused instances and processes that add to your cloud costs. It is amazing, right? Yes, it is a relief for all organizations.
If you are facing issues in any of the above cloud problems that we have mentioned, contact Conneqtion Group and we can assist you with your cloud problems. We can discuss Cloud adoption in detail and guide you through the benefits of Cloud automation for your business. Please feel free to contact us at firstname.lastname@example.org. If you have any comments regarding our work or would like us to include anything to this article, please let us know in the comments below.
Karan works as the Delivery Head at Conneqtion Group, a Oracle iPaaS and Process Automation company. He has an extensive experience with various Banking and financial services, FMCG, Supply chain management & public sector clients. He has also led/been part of teams in multitude of consulting engagements. He was part of Evosys and Oracle’s consulting team previously and worked for clients in NA, EMEA & APAC region.