October 2015 Summaries

Telegraf 0.2.0 has been released with several new features, including MQTT output, improved scalability through added "jitter" to flush periods, and a new agent configuration option `flush_interval`. The release also includes bug fixes and improvements to the collection interval, debugging, and test output. Additionally, the release notes mention changes to agent configuration values such as `precision` and `utcare`, which have been moved or removed. Telegraf's support for InfluxDB has also been enhanced with new plugins and features. The release is considered a significant update to the data collecting agent, offering improved performance and reliability.

InfluxDB Week in Review - October 26, 2015` highlights various news and updates from the InfluxDB community, including blog posts, videos, articles, and tutorials on using the platform. Several companies, such as Bleemeo, 5GTN, and Clustree, have adopted InfluxDB for its flexibility, scalability, and open-source nature. These organizations appreciate InfluxDB's ability to handle diverse metadata and tags, ease of integration with existing tools like Grafana, and fast data processing capabilities. The community also promotes user engagement through events, PRs for testing and benchmarking the new storage engine, and a call for company features on influxdb.com. Additionally, tech tips are shared, including support for long text data types and an asynchronous Scala client library.

Hacktoberfest is a month-long event that encourages people to contribute to open source projects by fixing bugs, creating new features, or updating documentation. DigitalOcean and GitHub sponsor the event, which offers free t-shirts to participants who open four pull requests on any GitHub-hosted project. InfluxDB is also participating in Hacktoberfest, offering its own t-shirt and stickers to developers who contribute to its codebase during October. To participate, developers simply need to submit their details via email after opening two accepted pull requests against InfluxDB or Telegraf.

Continuous queries in InfluxDB are pre-computed and stored aggregate query results that can significantly speed up summary queries without overloading the database. They are conceptually similar to materialized views, where expensive query results are pre-computed and stored instead of being computed on-the-fly. Continuous queries can be used to downsample data, transform and isolate data from one series into another, and "rollup" time-series data by time period. They can improve the end user experience for users in dashboard applications, enhance data analysis, and provide a more efficient way to access high-resolution data. To create continuous queries, developers need to run the `CREATE CONTINUOUS QUERY` statement against the query endpoint of the server using tools like curl. Once created, continuous queries are persisted as part of the metadata for the cluster and run periodically as data is collected.

We test InfluxDB's storage engines using various tools and techniques to ensure they work reliably and efficiently. We use CircleCI for unit-level testing, and more sophisticated testing as we approach the 1.0 release. Our test infrastructure includes collectd, Telegraf, and another InfluxDB system to store metric data. We record metrics such as disk IO, memory usage, CPU load, and resident set size during tests. A recent test ran for 8 hours on the new tsm1 engine, which showed steady write and CPU loads, with disk usage reaching a steady-state. The software does not suffer from detectable memory leaks. We run tests on various systems, including AWS EC2 instances, Digital Ocean droplets, and physical machines. Our testing strategies include stress testing and "burn-in" testing to ensure the software can handle long-term loads without issues. Future plans include more extensive testing for features like clustering and query performance.

InfluxDB has been featured in several recent articles, blogs, and videos that highlight its capabilities and adoption by various companies. GeoGraphite, Smarsh, and Row44 are among the notable users who have leveraged InfluxDB for their time-series data needs. These organizations appreciate InfluxDB's ease of deployment, statistical analysis capabilities, and compatibility with existing graphing tools like Grafana. Additionally, InfluxDB has been featured in several upcoming events, including meetups and webinars, which aim to introduce users to the platform and its applications. The community is also actively engaged through various channels, including a meetup program that rewards contributors with free t-shirts and other incentives.

The InfluxDB Week in Review highlights the latest news, updates, and achievements from the InfluxDB community, including new features, user testimonials, and upcoming events. The review covers various aspects of InfluxDB, such as its storage engine, query language, and HTTP API. It also mentions user success stories, including Mozilla's Raptor project and AXA Group, which use InfluxDB to store metrics and track performance statistics. Additionally, the review announces a new storage engine, provides tips for using InfluxDB, and offers discounts and freebies to users who participate in testing and benchmarking the new engine. The review concludes with an invitation for companies to feature their apps or products on influxdb.com and receive a thank-you gift, including an InfluxDB hoodie and stickers.

The InfluxDB team has developed a new storage engine called Time Structured Merge Tree (TSM Tree) designed specifically for time series data, aiming to improve disk space usage and query performance. The TSM Tree uses columnar storage and lifts the limitation on the number of fields in a measurement, allowing for efficient updates and compression techniques that adapt to different data types and precision levels. Similarities with LSM Trees are evident, but with improvements such as better compression and reduced disk space usage, resulting in up to 98% less storage space needed compared to previous versions. The new engine is currently available in nightly builds for testing purposes only and requires careful planning to test without disrupting existing data. Extensive testing and community review are encouraged, with detailed write-ups and instructions provided on how to get started.

The InfluxDB Week in Review, covering blogs, articles, how-tos, events, and testimonials from users such as Mirantis, Datacentred, and CATS Software, highlights the adoption of InfluxDB by various companies for its ease of deployment, performance, and features. Mirantis uses InfluxDB to store OpenStack infrastructure data, while Datacentred migrates their performance monitoring infrastructure from Graphite to InfluxDB. CATS Software finds InfluxDB's query language and API compelling, making it easy to monitor instance metrics and providing a simple web interface. The review also includes tips for accessing InfluxDB logs on CentOS and using an SNMP collector for InfluxDB 0.9.x.

The Chronograf team has released version 0.2.0 with several significant improvements, including embedding graphs in iframes for easy sharing and customization, support for multiple InfluxDB instances, custom time ranges, a persistence layer for configuration, and a native build for OSX. These changes enhance the visualization experience for InfluxDB users, making it easier to design and share graphs, manage multiple InfluxDB servers, and customize graph settings. The new release is designed to be more flexible and user-friendly, with features like customizable time ranges and improved distribution methods for different operating systems.