April 2024 Summaries

Vantage has introduced API service tokens, enabling users with owner-level permissions to create tokens associated with an account rather than a specific user, which enhances the management of automated workflows and resolves issues related to personnel changes and SSO integrations. These service tokens can be scoped for read and write permissions and can be attached to specific teams in Enterprise accounts, inheriting the team's role-based access control (RBAC) permissions, while non-enterprise users do not have this feature. This update replaces the previous system where API tokens were tied to individual users, which limited organizational control over permissions and posed challenges when users left the company. The service tokens also ensure that resources created with them, such as Cost Reports or dashboards, display the team associated with the token as their creator. The feature is free for all users, and there is no limit to the number of tokens that can be created, allowing for flexible and secure integration with services like the Vantage Kubernetes agent and Terraform.

Vantage has announced the release of new API endpoints, Terraform resources, and data sources to enhance the automation of cloud cost reporting workflows. These updates, available to all users including those on the free tier, aim to align the Vantage API and Terraform provider with the capabilities of the Vantage UI by enabling users to programmatically create, update, and retrieve resources such as anomaly alerts, virtual tags, and budgets. The new API endpoints and Terraform resources facilitate the integration of Vantage into infrastructure as code environments, allowing users to streamline cost management processes. Authentication is managed via Vantage API tokens, and future updates will continue to expand available features. Comprehensive documentation is provided to assist users in implementing these tools.

The blog post examines the differences and distribution of Intel, AMD, and Graviton processors in Amazon EC2 instances, considering factors such as performance, cost, and architecture. While Intel has traditionally dominated the EC2 landscape, AMD and Graviton are gaining traction, with Graviton offering significant cost advantages due to lower power consumption and being owned by Amazon. The x86-based architecture of Intel and AMD ensures broad software compatibility and supports multithreading, which is beneficial for certain workloads, whereas Graviton's Arm-based architecture may face some compatibility issues but provides cost-effective performance for specific use cases. The post highlights an increasing trend in the adoption of AMD and Graviton processors, reflecting a shift in customer preferences driven by workload-specific requirements and cost considerations. As of Q1 2024, the share of AMD and Graviton in EC2 spending has nearly doubled compared to the previous year, although Intel remains the processor for a larger variety of instance types.

FinOps as Code provides a framework for automating and visualizing unit economics by integrating business metrics with cloud cost data, enabling organizations to better align cloud spending with business objectives. The process involves creating a pipeline that uses a Python script to query a PostgreSQL database for business metrics, such as total orders, and then updates these metrics daily via the Vantage API. This setup allows e-commerce companies to correlate cloud infrastructure costs with specific business activities, offering insights into cost fluctuations and their impact on per-unit costs. By scheduling the script with tools like cron jobs or Apache Airflow, businesses can maintain an ongoing analysis of their cloud expenditure in relation to order metrics, as demonstrated by the example of fluctuating costs on weekends and specific dates. The approach, detailed in the FinOps as Code demo repository, requires familiarity with SQL databases, Python, and the Vantage API, and emphasizes the value of understanding cloud costs in the context of unit economics for improved operational efficiency.

Vantage has announced a closed beta for its Reseller and Managed Service Provider (MSP) Program, allowing these entities to offer cloud cost optimization services to their clients using Vantage's platform. This initiative is particularly aimed at resellers and MSPs who manage cloud infrastructure spending on behalf of multiple organizations through consolidated billing. The program is designed to address challenges such as detailed reporting and access to cost information, and it will initially prioritize support for AWS, with future plans for Azure and GCP. Participants in the closed beta will receive centralized management views and dedicated environments for each client organization. Vantage is engaging with interested MSPs and resellers to gradually expand the program, while also working on finalizing an annualized pricing model.

Vantage has introduced new visualization options for its Cost Reports, expanding the capabilities for users to analyze and interpret their cloud cost data. The update includes the addition of pie charts, which allow users to visualize the overall percentage of costs per service or grouping, providing a clearer high-level view of cost allocation. Furthermore, users can now access expanded date binning options for line and area charts, enabling daily, weekly, and monthly breakdowns alongside the traditional cumulative view. Multi-line charts have also been introduced, allowing for grouped cost data representation, where each line corresponds to specific categories such as service or resource. These enhancements are available to all users at no additional cost, enhancing the flexibility and depth of data analysis within the Vantage platform.

In a detailed comparison between Anthropic's Claude models and OpenAI's GPT models, the blog discusses the advancements and pricing structures of these AI language models as of mid-2024. Claude has emerged as a strong competitor, particularly with its Claude 3.5 Sonnet model, which has shown significant improvements in benchmarks and functionality, excelling in tasks like writing, summarizing, and coding. Claude's models offer a substantial 200k token limit, surpassing OpenAI's highest at 128k tokens, and they are accessible through Amazon Bedrock with two pricing options: On-Demand and Provisioned Throughput. Meanwhile, OpenAI's GPT-4o model remains widely regarded for its advanced capabilities, available through Azure with a straightforward pay-as-you-go pricing model. The blog highlights that Claude 3.5 Sonnet is more cost-effective than its predecessors and competitive with GPT-4o, particularly in terms of input token pricing, while noting that the choice between these models largely depends on specific needs such as cost, availability, and token limits.

Vantage has introduced a feature allowing users to project costs up to six months ahead in their Cost Reports, enhancing the previous limitation of only forecasting for the current month. This update is particularly beneficial for users engaged in quarterly and yearly financial planning, as it eliminates the need to export data to generate extended forecasts. By using a date picker in the Cost Reports screen, users can now select forecast periods ranging from the upcoming month to six months into the future. Enterprise customers with expanded retention periods can forecast for a correspondingly extended duration. The forecasts, generated through a time-series model that accounts for seasonality and anomalies, are available on bar and line charts for various date bins, although they are currently limited to individual services and overall data rather than filtered or grouped data. The feature, free for all users, is accessible immediately after saving a report and reflects in less than 10 minutes, although exporting these forecasts is not yet available.

Vantage has introduced Kubernetes rightsizing recommendations to help users optimize resource allocation and reduce costs by providing specific suggestions for adjusting CPU and RAM settings on managed workloads. The new feature is designed to minimize idle costs and prevent overprovisioning by offering detailed efficiency reports for clusters, namespaces, labels, and containers. Users can access these recommendations via the Cost Recommendations screen, where potential savings are highlighted, and the Rightsizing tab under each workload displays average, maximum, and recommended settings along with usage graphs. To utilize the recommendations, users must upgrade to Vantage Kubernetes agent version 1.0.24 or later, with recommendations appearing after 48 hours of data collection, although 30 days of data is advised before implementing changes. The rightsizing recommendations are available to all Vantage users, including those on the free tier, and support major cloud providers like AWS, Azure, and GCP.

Vantage has introduced a new feature called virtual tagging, which allows users to create custom cost allocation tags for cost data imported into the platform, addressing issues with inconsistent tagging practices across different cloud providers. This feature enables users to apply tags to providers that do not support them, enhance tagging coverage for those that do, and resolve tagging discrepancies, thus facilitating more accurate cost reporting. Users can define virtual tag keys and values with filters directly within Vantage, circumventing the need for engineering teams to alter cloud infrastructure, and these tags can then be used for various functionalities such as cost reports and allocations across the platform. The feature is freely available to all Vantage users, and it supports all provider integrations within Vantage, even those that do not natively support tags.

Vantage has introduced a feature that allows users to calculate unit costs using metrics imported from Datadog, a popular metrics system, enabling visualization of metrics like cost per user or cost per request on Vantage's Cost Reports. This addition simplifies the process for users who already rely on Datadog by providing a direct import option, eliminating the need for additional processes or integrations. Users can now create new business metrics in Vantage's Financial Planning section by selecting Datadog as the import method, allowing for metrics to be automatically imported, initially covering the past six months and updating daily. This feature is beneficial for those looking to analyze metrics such as requests per second or daily active users, and is available for all users at no additional cost. The integration requires specific permissions and is designed to seamlessly work with existing Datadog integrations in Vantage, ensuring users can select appropriate metrics and scales for their cost analyses.