We are proud to present the latest release of StackState!
This release is yet another significant step towards completing our vision toward fully autonomous operations using the 4T Data Model.
Broadly speaking with this release we are making strides in four areas:
- User experience - The 4.0 release was the first release that contained "Perspectives". In this release we've enhanced the navigation between and over those Perspectives to find the root cause of issues lightning fast. A Perspective is a way of looking at the 4T Data Model. Think of each T as a dimension, the entire space being your IT, spanning across infrastructure, applications and business. A view in StackState is a looking glass in the form of a four-dimensional cube. Finding the root cause of an issue is a process of elimination that narrows the volume within the cube. Which side of the cube you are looking at is determined by your perspective. Finding something in the 4T Data Model is a matter of shrinking (narrowing the view) and rotating the cube (changing the perspective).
- Intelligence - This release marks the birth of our fully autonomous AI. AI features of StackState thus far required some form of configuration, which requires both data science and domain knowledge, something not too many companies have. Our new autonomous AI means zero configuration is needed. You turn it on and it will start producing meaningful results that help you achieve your goals.
- Data quality - With the 4.0 release, StackState already had the ability to process distributed traces from multiple sources. With the 4.1 release, this capability is extended with the addition of a Trace Perspective, which helps you quickly find slow or erroneous parts of your (distributed) system.
- Scalability - StackState is built for large enterprises, where all the data is already present, but the understanding is lacking. We've chosen to use Kubernetes as our deployment platform. This enables StackState to run with ease both on-premise and in the cloud at scale, self managed or managed by us. This release we're removing the beta label of the StackState Kubernetes deployment.
Traces, also referred to as transactions, reveal how requests to various services in IT environment are passed down from service to service until the request is fulfilled or an error occurs. Traces are useful for identifying performance bottlenecks and understanding how the different types of traffic are affecting distributed systems.
StackState previously supported traces as a source of topology and telemetry, but the traces themselves were not exposed. We've added Trace Perspective to help you identify root causes of down-time and performance issues even faster.
Traces are gathered by the StackState trace agent for Java or .Net services or retrieved from existing trace sources such as AWS X-Ray and Traefik. To get started with any of these sources, go to the StackPacks page and select "Tracing" from the tags.
Autonomous Anomaly Detector - Beta
Anomalies are a way of identifying abnormal behavior and are especially useful for:
- Assisting root cause analysis.
- Identifying interesting behavior.
- Alerting on abnormal behavior.
StackState already offers anomaly detection based on metric stream baselining, which requires a number of configuration parameters per metric stream type (see the docs). This puts a burden on the user to know both how to configure an anomaly detector based on (fairly complicated) algorithms and to understand the data. The Autonomous Anomaly Detector is StackState's first fully autonomous AI feature, thus completely removing this burden.
Autonomous = Zero configuration. Once you've installed the Autonomous Anomaly Detection StackPack anomalies will automatically start flowing in.
Autonomous Anomaly Detection scales to large environments by prioritizing anomaly detection on metric streams based on its knowledge of the IT environment and user preferences. Streams with higher priority are given more attention by the Autonomous Anomaly Detector, thus effectively balancing the amount of resources required by the anomaly detection algorithms and the value that is produced by doing anomaly detection. Auto Machine Learning (AutoML) is used to determine how to best approach anomaly detection for a set of streams.
(screenshot of anomaly events founds in the Events Perspective)
The Autonomous Anomaly Detector StackPack is generally available for everyone as part of the 4.1 release but still has a beta label on it. We're starting a beta program with our customers to further tune the machine learning algorithms and models used by the Autonomous Anomaly Detector so as to minimize false positives and maximize true positives in real production environments. If you would like to join please contact us.
The Autonomous Anomaly Detector requires that StackState is deployed on Kubernetes.
StackState on Kubernetes
StackState can now be run on Kubernetes. This allows you to run a highly available version of StackState on-premise or in the cloud, using the same underlying infrastructure technology. Installation is performed using Helm, the standard package manager for Kubernetes.
Supported Kubernetes platforms include OpenShift, AKS (Azure Kubernetes), or EKS (Amazon Kubernetes).
Starting with 4.1.0, StackState on Kubernetes is the preferred method of installation for StackState. The following features are only available when running StackState on Kubernetes:
- High availability with auto fail-over.
- Autonomous Anomaly Detection.
Deprecation schedule for Linux packages
For this release, StackState can still be deployed using Linux packages. However, please be aware of the deprecation schedule for Linux packages:
- For the next major release, tentatively named 4.2.0 and scheduled for December 2020, Linux packages will be deprecated
- For the following major release, tentatively named 5.0.0 and scheduled for March 2021, Linux packages will no longer be made available.
- Product releases are supported for the duration of the next two major releases. That means Linux packages for 4.2.0 will be supported until at least June 2021.
Many other improvements
- Global 4T navigation - Regardless of whether you are in the topology, telemetry, events, or the newly released Traces Perspective you can now filter down on all dimensions with the same set of filters and easily change perspectives.
- Starred Views - It is now possible to create favorite views by starring them. Starred Views appear as shortcuts in your main menu.
- A binary version of the StackState CLI is now available for both Windows and Linux.
- Agent integration monitoring - StackState agent instances appear automatically as components inside StackState, including a check and event stream that makes it easy to monitor their performance.
- Many new integrations, including support for Docker Swarm, Nagios ITRS OP5, and Humio.
- Many UX improvements - We've listened to your input and used it to improve many parts of the user interface experience.
- Many performance enhancements - The overall responsiveness of the product experience and the amount of 4T data StackState can handle has improved. Performance improvements were mostly focussed on topology synchronization and on the speed of state changes in checks and views.
StackState offers technical support for the three most recent major releases. The release of StackState 4.1 marks the End of Life (EOL) of the StackState 1.14 version range. Consequently, this version of the product will no longer be supported and we encourage customers still running the 1.14 version range to upgrade to a more recent release.
Upgrading to 4.1.0
Before upgrading to StackState 4.1.0, please review the upgrade instructions provided in the documentation.
Configuration changes in 4.1.0
- In this release the
sts-healthurihas been moved from port 7071 to 7080 in
processmanager.conf. Custom made
processmanager.confconfigurations will need to be adapted.
You can download the release at the StackState Software Distribution site.