August 2024 Summaries

Vantage has introduced support for Custom Providers, enabling customers to upload cost data from sources not previously integrated with Vantage, such as providers lacking billing APIs or custom internal systems, using the FinOps FOCUS schema. This feature is designed to offer a standardized approach to tracking expenses from various SaaS vendors, which often lack consistent billing support, through CSV uploads or API posts, allowing for a cohesive view of expenses within an organization. The Custom Providers integration is included under the standard Vantage pricing model and is accessible to all Vantage customers, providing them with the flexibility to manage both one-time and ongoing costs. This initiative aims to centralize cost reporting and allocation by using the FinOps FOCUS standard, which is gaining traction as a universal schema for usage reports, ensuring compatibility with Vantage’s existing features like anomaly alerts and forecasting.

Vantage, a cloud cost visibility and optimization platform, used its own services to achieve significant cost savings on AWS by identifying anomalies and optimizing resources. They switched from costly NAT gateways to a combination of NAT instances and VPC endpoints, reducing external traffic costs by 80%. This was facilitated by Network Flow Reports, which pinpointed expensive data ingress and egress sources. In addition, Vantage tackled rising S3 storage costs by implementing Amazon S3 Storage Lens and lifecycle rules, effectively managing over 500 TB of temporary data and cutting storage usage in half. These efforts underscore the platform’s effectiveness in streamlining cloud cost management and demonstrate how Vantage’s tools can be harnessed for both internal and external cost-saving strategies.

Vantage has introduced Autopilot for AWS Savings Plans, a feature that provides automatic purchase recommendations and transactions of Compute Savings Plans to optimize spending on AWS services like EC2, Lambda, and Fargate. This development comes in response to AWS policy changes that have complicated the management of Reserved Instances (RIs), prompting Vantage to phase out RI support in favor of the more flexible Compute Savings Plans, which are not tied to specific regions or instance types. Customers can configure their Savings Plan settings with various term lengths and payment plans, and Autopilot will handle ongoing monitoring and application of these plans while charging a 5% fee on savings realized. Existing Vantage customers with RIs are encouraged to transition to the Savings Plan model, allowing Autopilot to manage their commitments and list current RIs for sale. The new system aims for optimal coverage and offers a dashboard for monitoring savings and commitment levels. The initiative reflects a strategic shift to accommodate AWS's evolving policies and enhance customer flexibility and cost efficiency.

Vantage has introduced a new feature allowing both teams and individual users to set default dashboards, which are the initial view upon logging into the platform. Previously, all users were directed to a generic Overview screen that often contained irrelevant information, prompting them to navigate to the dashboards they needed. Now, users can immediately access dashboards tailored to their needs, improving efficiency and user experience. This feature is available at no extra cost but requires an Enterprise-tier subscription for team-level defaults. Owners can configure these defaults through the Team tab in the console's settings, while individual users can set personal defaults in their profile settings. If both personal and team defaults are set, the personal dashboard takes precedence, and for users on multiple teams, the default is determined by the order of team addition.

Gemini 1.5 Flash emerges as a leading small language model due to its affordability and extensive one-million-token context window, making it ideal for cost-sensitive applications. In the competitive landscape with OpenAI's GPT 4-o Mini and Anthropic's Claude 3 Haiku, each model presents unique strengths for different use cases. GPT 4-o Mini, released in July 2024, offers efficient performance for chaining and parallel model calls, particularly in processing large contexts and real-time support applications. Claude 3 Haiku is recognized for its speed and text processing capabilities, suitable for tasks like content moderation and data extraction from unstructured sources. Despite differences in pricing and performance metrics, such as MMLU scores and token capacities, Gemini 1.5 Flash stands out with its cost-effective pricing and expansive context capabilities, supporting tasks that range from object recognition to complex reasoning.