Forecasting and Anomaly Detection are two features in Octo that help organizations to optimize resource utilization and allocation effectively. By using these features, you can keep an eye on and understand your future cloud cost trends, and get alerts every time an unusual cloud cost spike happens.

Being able to predict and track your future cloud spending accurately is super important for planning ahead, setting budgets, and quickly reacting to avoid cloud costs from going crazy. Plus, spotting any weird expenses early on can stop you from getting unexpectedly high bills. These features will help you pinpoint the causes of unexpected costs, enabling you to take corrective actions and avoid repeating the same costly mistakes.

Why is it incorporated in your Cost Group page?

As mentioned in the Cost Groups article, your cost groups form the foundation of all data displayed in Octo, making it the best spot to handle your cloud expenses smartly. This is also the reason why every cost group that you create will be incorporated with these features. By checking, organizing, and keeping an eye on all the important details using your cost groups, you can use Forecasting and Anomaly Detection to keep your cloud spending in check all the time.

Forecasting, in general terms, is the practice of predicting future cloud spending. This is typically done by analyzing historical spending data and assessing projected costs for upcoming projects or events. According to Forbes, forecasting involves using past data and trends to estimate future expenditures, which is crucial for managing financial expectations.

In Octo, Forecasting is a feature designed to give the users advantage by providing estimates of possible expenses over the next few days, based on the specified date range. It allows businesses to anticipate expenses and avoid financial surprises. This feature is particularly important for budget and planning as it helps users understand how their expenses will evolve over time.

The Forecast feature in Octo leverages machine learning, specifically ARIMA (AutoRegressive Integrated Moving Average) Plus in Google Cloud Platform’s BigQuery, to predict future cloud costs. This approach allows the system to forecast up to one year of data, including the current month.

To generate your forecasted data, you can try following these steps:

An anomaly, in general standpoint, is simply something that deviates from what is considered normal or expected - a deviation from the usual pattern or trend. May it be in weather, finance, health, or data, anomalies can occur in any given situation. In the context of cloud cost management, however, an anomaly is defined as a situation where the total cost is either significantly above or below a specified threshold. These deviations from the expected cost pattern trigger alerts, prompting further investigation to see if resources are used efficiently and costs remain under control.

Anomaly Detection in Octo is pretty much the same. This feature allows users to identify unusual patterns or deviations in cloud costs that might indicate inefficiencies or potential issues. Integrated into every cost group you create, Anomaly Detection leverages artificial intelligence to detect possible anomalies based on your historical data.

Using Google Cloud Platform’s BigQuery and ARIMA Plus, Octo analyzes and trains on your past data to predict and identify anomalies in your cloud costs. This ensures that any unexpected spikes or drops in spending are quickly recognized, allowing you to take prompt action to manage your resources more effectively.

As we’ve established, anomalies are quite disturbing as they indicate that something unusual is happening with your cloud costs. The ideal approach to this is to identify these anomalies and determine their source. Then, you can make more informed decisions to effectively optimize your cloud costs.

To view anomalies in your selected cost group, navigate to the ‘Overview’ tab where your cost group chart is displayed. To activate the Anomaly Detection feature, click on the setting icon, located at the upper right corner of your screen - between the ‘Group by’ dropdown button and the reload icon. Once activated, the chart will display red marks that indicate any anomalies in your cloud costs.

These red marks highlight unusual spending patterns, allowing you to quickly identify and address potential issues.

In this example, only one anomaly is detected. Click the red mark to see details such as the date, account, service, and the cost associated with that specific anomaly. If you want to know more information about that anomaly, just click CHECK DETAILS OF ANOMALIES and it will take you to the anomaly tab.

All anomalies within the specified date range for your cost group will be displayed under the Anomaly tab. This tab provides more detailed information about each anomaly, what each column entails, and other small features within the tab. These are explained as follows:

Search Feature

This feature is especially helpful when you're dealing with a lot of data and it gets a bit overwhelming. You can just type in the account id, account name, or product that you want to view.

Filter Button

The main purpose of this feature is to filter specific data for display. But, how does it work? To begin with, we have three entities that can be set in the filter: minimum anomaly cost, accounts, and products. When it comes to account and product, it’s pretty straightforward and easy to understand - simply click on the account and product dropdown, apply selection, and the data will display. With Minimum Anomaly Cost, you will have to set a specific cost first. Octo will then show you only the accounts with anomaly cost within or higher than that specified amount. It’s like setting a minimum threshold of what you want to see.

Select Date Range

You can select the date from which you can see the dates when the anomaly happened. You can do it in the last seven days up to a month, year, or quarter in the quick select range if you don’t want to input the date manually.

Edit Information Setting

This feature allows you to display the anomaly data using the absolute value of the anomaly cost or percentage of the total cost. This way you can customize your Anomaly tab when it would retrieve data based on the threshold that you set. You can also choose up to nine months of past data to use as historical data for detecting possible anomalies.

Anomaly detection in Octo is easy to understand. For example, if you want to detect the possible anomalies for this month - the data nine months backward will be used to train the machine learning model and detect the anomaly. Data to be confirmed: This model is updated daily at 7 PM. Keep in mind that the data in the Anomaly tab is not finalized until the end of the month, so you might see some changes daily. This is because the Cost Usage Report (CUR) is constantly updating, and there are times when there's no data available for certain days. It can take a few days for new data to be added.

Since CUR data are used to predict anomalies, thus the anomaly table will also be affected and will be constantly changing. It will take up to a month for these data to be stable.

Two key variables are essential in Octo's Anomaly Detection feature: the lower limit and the upper limit. These thresholds help determine what's considered an anomaly and what's considered normal.

Think of it like a range:

Essentially, if your data falls outside this range, it's flagged as an anomaly, indicating a potential issue or unexpected change in your cloud usage.

Now that you understand the purpose and how to use these two features, it's time to learn what to do when Octo detects an anomaly in your cloud usage.

In our fast-paced industry, it's impossible to focus on just one task, especially if you're leading a team or running a company. You need to be adaptable. So, to be notified about any issues with your cloud costs even when you're busy with other things, the next step is to set up alerts.

Creating alerts in Octo is a breeze, especially if you're familiar with navigating its features. But don't worry, even if you're new to Octo, we've got you covered! Click the button below for a smooth and easy experience.