May 2022 Summaries

OpenTelemetry is a unified observability framework that provides a common interface for instrumenting code, collecting metrics, logs, and traces in a consistent manner. It differs from previous observability tools by focusing on the creation and transmission of signals, rather than storage or query capabilities. This allows developers to use OpenTelemetry as a single API and SDK per language, offering common concepts across metrics, logs, and traces. The framework enables stream processing, enrichment, aggregation, and routing of signals, making it easier to deploy and manage telemetry data. However, querying signals remains a challenge, with different query languages for each signal type (e.g., PromQL for metrics, LogQL for logs). A unified language that can handle multiple signal types is needed, which could be based on SQL, a well-established standard for modeling relationships and analyzing data. Adopting SQL as the query language for OpenTelemetry could provide a consistent approach to querying telemetry data across different signal types, enabling seamless correlation and analysis of metrics, logs, and traces.

This installment of the "Community Member Spotlight" series features Julien Salinas, full-stack developer, founder, and CTO at NLP Cloud, who discusses how his team is building an advanced API to perform natural language processing tasks while taking care of the complexity of AI infrastructures for developers. NLP Cloud is an advanced API for text understanding and generation in production, offering several API clients for easy integration with popular programming languages. The company uses TimescaleDB to track API usage, which provides continuous aggregates, automatic data downsampling, and a PostgreSQL ecosystem, making it easier to implement analytics solutions without learning new skills. Julien's team is currently working on multi-account capabilities and integrating new AI models, while also optimizing the speed of their Transformer-based models. The company recommends TimescaleDB for development teams looking for a robust and easy-to-deal-with time-series solution, and provides advice on implementing scalable database strategies by separating business databases from analytics databases.

The cryptocurrency market is evolving rapidly with over a thousand cryptocurrencies already in the market and an ICO almost daily. Data analysis using PostgreSQL and TimescaleDB can provide insights into this market, helping to understand its dynamics and risks at both micro-level (e.g., personal investments) and macro-level (e.g., prevention of market crashes and major loss of capital). Analyzing historical OHLCV price data on over 1200 cryptocurrencies reveals insights into the cryptocurrency market, such as the volatility of Bitcoin prices, the impact of Chinese regulations on trading volumes, and the impressive growth of Ethereum. The long tail of cryptocurrencies can be sifted through by transaction volume or profitability, with Litecoin, Ripple, and Ethereum Classic showing promise alongside Bitcoin and Ethereum. However, it's important to remember that the market will continue to change, and data analysis should be regularly updated for accurate insights.

Tracing is a key component of observability, capturing a connected view of how requests travel through an application and allowing developers to understand how different components interact with each other. To analyze trace data effectively, we need powerful tools that can handle complex analytics, such as SQL. Observability is essentially an analytics problem, and leveraging SQL for trace analysis can unlock new insights to troubleshoot production issues faster and proactively improve applications. Promscale is an observability backend built on PostgreSQL and TimescaleDB, providing full SQL superpowers to solve even the most complex issues in distributed systems.

This is a summary of OpenTelemetry metrics in 1 paragraph. OpenTelemetry is an open-source observability framework that simplifies the process of collecting signals from services and sending them to the backend of choice. It provides six instrument types: counters, asynchronous counters, UpDownCounters, asynchronous UpDownCounters, histograms, and asynchronous gauges. Each type has its own characteristics and use cases, such as synchronous vs asynchronous implementation, aggregation types, and attributes attachment methods. OpenTelemetry aims to define a single standard across all types of observability data, including metrics, logs, and traces. It provides a Collector that can be used to reaggregate and redirect metric data, and a global MeterProvider that creates a Meter and associates it with one or more Instruments. The framework allows for flexible transformation in the collection path, supporting multiple export types, including push and pull options. This enables interoperability between existing metric systems, making OpenTelemetry an attractive option for developers looking to standardize their observability data.

Database monitoring is crucial for effective data management and building high-performance applications. The `pg_stat_statements` view provides valuable information on query performance, but its cumulative nature can make it challenging to pinpoint problems. To overcome this limitation, storing the data in a time-series format with hypertables allows for efficient querying of trends and patterns over time. This enables the creation of continuous aggregates and data retention features, providing a complete solution for self-serve query monitoring. By leveraging TimescaleDB's features, such as compression, automated job scheduling, and data management, organizations can establish a robust database monitoring system that supports their needs for scalability and performance.

This month's newsletter from TimescaleDB highlights various product updates, technical content, videos, tutorials, community events, and interviews. The company released TimescaleDB 2.6.1 with several bug fixes, and announced its introduction to Timescale Live, live tech talks for developers building data-intensive applications. Other notable events include o11yfest 2022, PostgreSQL Conference Germany, Data on Kubernetes Community Day 2022, PrometheusDay Europe 2022, GlueCon 2022, KubeCon Europe, Write the Docs Portland 2022, and PGCon 2022, showcasing various opportunities for developers to learn, share, and network. Additionally, TimescaleDB shares tips, reading lists, and more in-depth content on topics such as Prometheus metrics, continuous aggregations, and PostgreSQL performance bottlenecks.