Are you leaving money on the table?

In the “FinOps for Enabling Value in Cloud” Dec 2023 report, Everest Group listed the following as key cloud-value related challenges.

·          No insights from the available cloud landscape data

·          Unplanned wastage and inefficient sizing of cloud resources

·          Increase in cloud operation costs

·          Inadequate operational performance

·          Inefficient collaboration between internal teams

·          Difficulty in managing governance, risk and compliance

And following on that, the 2024 Key Issues Report (Everest Group), lists cost optimization, revenue growth and customer experience improvement (linked to operational performance) to be the top priorities going into 2024.

While the key cloud-value related challenges listed above cover the whole nine yards, there are at least four aspects that are very relevant to the topic of discussion:

·      Unplanned wastage

·      Inefficient sizing of cloud resources

·      Increase in cloud operation costs

·      Inadequate operational performance

The 2024 Flexera State of the Cloud report found that 27% of the cloud spend is wasted and continues to be the top cloud challenge.  The right balance is about delivering on the business outcomes without overspending. And that’s done through cloud optimization.

While typical cost optimizers (both native and third party) deal with reducing unplanned wastage and hence reduction of costs, the USP of IBM Turbonomic is of doing the same while guaranteeing application performance, by continuous resource management.

Capacity Planning Vs Capacity Management

Let’s examine two terms: capacity planning and capacity management.

Capacity planning: determine future resource requirements of an IT infrastructure to ensure that it can meet the anticipated demands of the organization. It involves forecasting future workloads, analyzing historical data, and making informed decisions about resource allocation.

Capacity management: continuously monitor, analyze, and optimize resource utilization in real-time. It involves the proactive management of resources to ensure that the infrastructure is effectively utilized, performance targets are met, and costs are optimized.

What's your choice?

Active management solutions like the IBM Turbonomic platform are geared for modern environments. They leverage automation and enable real-time adjustments that help reduce cost and improve the availability and performance of cloud environments and applications. Continuous optimization enables a Turbonomic customer to exploit on frequently updated CSP offerings.

Innovation: It’s everywhere, but is your environment ready?

A good innovation should always be about passing more benefits to the customer. Now the big questions: As a customer, are you equipped enough to absorb the benefits of constant innovation? Are you leaving money on the table?

This is a broad topic, but we will limit the exercise to the offerings of the Hyperscalers. All the Hyperscalers provide many capabilities, including IaaS, PaaS, SaaS and many more. The illustration below is going to be on the Compute Offerings of AWS and Azure, with its multitudinous variants and hence the challenge it throws for the Cloud administrators. And eventually bloating the cloud bill.

Figure 1 Partial list of AWS EC2 offers on AWS Marketplace

As you can see, innovation is constant. Improvising from T1 to T3 (incorporating new processor types, various instance types, improving CPU credit systems, better price points and availability across regions and more).

To add to the complexity, let's delve a little bit into the various instance types within T1, T2 and T3.

Figure 2 Not meant to be an exhaustive or accurate information on EC2 offerings. Check AWS Marketplace

So, what’s the complexity here? We can apply a bit of elementary math to explain this.

Figure 3 Is it humanely possible to choose the right destination VM type for every one of your 1000’s of VMs?

The number of combinations, as shown above, does not include the following:

1. Number of public clouds you have your VM's on (More than 80% customers are on multiple clouds)
2. Enterprise Agreements, Savings Plans, Reserved Instances
3. AWS Auto Scaling groups, Azure Scale sets

IBM Turbonomic's secret sauce is ensuring that all the decisions are about guaranteeing application performance, by defining a desired state (SLO). How do you factor this in the formula above?

A few questions to consider

1. What is the size of your VM deployment in cloud or in the data center? Is it in tens, hundreds or thousands?
2. Are you leveraging one public cloud, multi-clouds or in a hybrid cloud model?
3. Are you sure that your VM instances are of the right type, as you deploy?
4. How often do you reassess if the applications are running on the right instance type?
5. Are you sure that you are using the innovations provided by the CSPs?
6. And most importantly, how are you ensuring that the applications running on the VMs has the right resources allocated to them, at the right cost?

Consider IBM Turbonomic

We asked few questions, and we will see if IBM Turbonomic has an answer to all of them. The following are direct screen captures from IBM Turbonomic.

1.             What is the size of your VM deployment in cloud or on-prem? Is it in tens, hundreds or thousands?

2.             Are you leveraging a single-cloud or multiple clouds, or in a hybrid cloud model?

3.             Are you sure that your VM instances are optimal as you deploy?

4.             How often do you reassess your instances? Are the applications running on the right instance type?

5.             Are you sure that you are using the innovations provided by the CSPs?

And most importantly, how are you ensuring that the applications running on the VMs has the right resources allocated to them, at the right cost?

1.            In the picture above, IBM Turbonomic has identified 82 scaling actions, under the action category “Savings”.

2.    “Savings” are about ensuring efficient utilization of resources and thus bringing down your bill. You can see that the actions are about moving from one Instance type to another, with the cost impact shown. Remember the huge combinations and permutations that was discussed earlier. IBM Turbonomic does all those, without affecting the performance. Let’s detail on the types of savings suggested here.

a.    The top three savings in the picture above, is about moving to an Instance type (m5a.4xlarge to m5.4xlarge), to utilize the Discount coverage. (Clicking on the Details tab, will provide information on why this action is being proposed. Will be discussed below). And you could see that this action saves $444.78/month.

b.    The suggestion to move from c3.2xlarge to m5a.xlarge is owing to “Underutilized VCPU” and thus saving $215.92/month.

So, what’s in that “Details” tab? Let’s explore the topmost action in the previous image.

The above action is about reducing your cloud costs, by recommending scaling VMs to instance types that are changed discounted rates. The chart above shows how well IBM Turbonomic has ensured to utilize your current discount inventory, from a current level of 0% to 100%, post the actions. The desired goal is to maximize the utilization of your inventory and thus take full advantage of the discounted pricing offered by your cloud provider.

And what about the “Performance Actions” … Here you go!

• In the picture above, IBM Turbonomic has identified 39 scaling actions, under the action category “Performance”.
•       Performance actions are all about actions when IBM Turbonomic detects conditions like CPU congestions, Memory congestions and IOPS needs, that directly puts your application performance at risk.
• .     Let’s walk-through one of the actions by clicking the details button.

What do the details tell as in the picture above?

1. VMem congestion has been identified as the cause for the performance issue, owing to which a switch from Standard_F2s_v2 to Standard_D2as_v4 is suggested.
2. This also means the spending will increase by $3.58/month
3. VMem percentile dashboard gives the information on the diagnosis. vMem utilization has hit 100% a few times. And after taking actions (Blue dotted line), IBM Turbonomic has analyzed that the vMem utilization will be 50%, thus resolving the congestion. The same is illustrated in the table next to the chart.

In the section above, We discussed scaling actions recommended by Turbonomic, one aimed at improving efficiency and the other at enhancing performance. Now are these the only types of actions recommended. Definitely not. Please find a comprehensive list of actions and action categories that are recommended by IBM Turbonomic.

Turbonomic generates actions based on how entity types use or provide resources, and what each entity type supports. Please consult Turbonomic documentation on “Actions by Entity Type”

Turbonomic is the premier solution for Application Resource Management (ARM), a hierarchical, application-driven approach that continuously analyses application’s resource needs and generates fully automatable actions to ensure applications always get what they need to perform. It runs 24/7/365 and scales with the largest, most complex environments. IBM Turbonomic is your trusted partner, that ensures that you take advantage of the Hyperscaler’s innovations.

You can download a 30-day free trial today or meet with an expert and discover how Turbonomic can help continuously assure application performance at the optimal cost. Visit IBM.com/Turbonomic today and get started.

Author:

Veeramani Nambi, Product Manager - Turbonomic Software

d