Most organizations regularly experience cloud spend anomalies due to inadvertent internal misconfiguration issues. These internal misconfigurations can be related to provisioning and leaving idle large instances for evaluation purpose, large amounts of data transfer out or frozen serverless calls. Often, the responsible engineers are looking to experiment with newer cloud-native technologies for app development, building data lakes or testing load and scale performance of an application by spinning up instances or services. A simple misconfiguration can spin up the wrong number of instances or services, or cause a massive data transfer, causing spikes in the bill which can go unnoticed for a long period of time but can cost the company considerably.
In addition to the typical unintentional spending anomalies, some are malicious. Recently Tesla was in the news – no, not for introducing new car models, instead it was for hackers using Tesla’s public cloud for mining cryptocurrency – called “cryptojacking.” Cryptojacking is a technique hackers use to leverage public cloud infrastructure that has been compromised with the goal of mining cryptocurrency.
Detect and Correct Anomalies Before They Wreak Havoc
Whether intentional or unintentional, spending anomalies are often not detected early and impact a company’s bottom line. At Cloudability we’ve helped customers manage billions of dollars of cloud spend and observed that a lot of our customers were being impacted by a lack of understanding of their cloud bill and its ramifications. To solve these problems, Cloudability is pleased to announce the launch of our Anomaly Detection capability.
You might ask, what’s so unique about Cloudability’s Anomaly Detection? Let’s dig deeper into why Cloudability’s Anomaly Detection can provide the unique insights and correlation that are hard to replicate across other industry platforms.
Cloudability is a True Cost™ cloud management platform that stores customer data perpetually with no loss of fidelity. This data lake gives Cloudability the ability to deliver deep insights and visibility to the customer. Cloudability’s Data Science team has built machine learning algorithms that continuously analyze data and alert the user when anomalies are detected.
When evaluating an anomaly detection solution and the accuracy it delivers, the following aspects should be taken into consideration:
- Richness and quality of data set
Models for machine learning algorithms are only as good as the training data available. Since Cloudability’s data lake has rich training data, it ensures improved machine learning algorithms and accurate results. Cloudability’s ability to store customer data in perpetuity without loss of fidelity ensures high-quality data for analysis.
- Accuracy of data
Accurate data is at the heart of detecting and solving problems. Inaccurate data can create false positives and mislead organizations – breaking trust in a solution, ultimately leading operators to ignore alerts and exposing themselves to bigger potential problems. In the case of cloud spend management, it is essential that a solution have accurate data. Inaccurate data creates too many false positives and ignores conditions that should trigger alerts. Cloudability’s True Cost™ cloud management platform delivers the true cost of cloud infrastructure, since it includes discounts, amortization and credits extended by the cloud providers.
- Timeliness of data
In the case of anomalies, it is important that the organizations know about a problem as early as possible. Detecting problems early is critical to limiting the scope of a breach or reducing the unwarranted impact of an error. Cloudability’s platform is the industry leader in time to process cloud spend and usage data, ensuring that any anomaly detected in a customer account will trigger a notification within a few hours so the problem can be mitigated before there is a significant negative impact to the organization.
Cloudability’s Anomaly Detection is enabled for all customers, and starts working right out of the box. Users don’t have to fine-tune model parameters or figure out which algorithms would work best. Cloudability’s True Cost™ platform detects and delivers insights not possible manually.
Cloudability processes anomaly data as soon as a new billing file is available from the cloud provider. This ensures that anomalies are detected quickly and can be addressed. Cloudability processes billing data from multiple cloud providers and presents all anomalies in a single pane of glass.
Cloud providers recommend using tags as a best practice for cloud operations. In the case of anomalies, proper tagging helps Cloudability provide deeper insights and correlate anomalies to the service(s) or instance(s) being impacted.
Users don’t have to define any preconfigured settings to receive alerts. However, Cloudability does provide users the ability to set up email alerts for when the unusual spend goes over a certain amount and receive them on the go.