The Craig punchplatform version 5.0.0 is released. This post gives an overview of the new features.
The Craig punchplatform has a number of important new features.
- Elastic 6.x stack
- Kibana 6.2.2
- support for apache avro encoding and type conversions
- New administration REST and HMI platform architecture and tools
- Integrated central log and alert management
- Punchplatform Marchine Learning official GA release
The Elastic 6.x stack is a significant move forward. Checkout the elastic blog for details.
“An entirely new zero-downtime upgrade experience, the addition of fast, op-based recovery thanks to sequence IDs, improved handling of sparse data, faster query times, distributed Watch execution, and the list keeps going. The summary, or aggregation, of features is in the detail post“. Some new important features are :
- major improvements for sparsely populated features
- new types and search options (mostly enabled by the update to Lucene v7):
- ip_range to deal efficiently with ipv4 or ipv6 address ranges
- significant disk space savings in many out-of-the-box situations
- index sorting to ensure faster search
- improved hard recovery
The punchplatform integrates and relies on the elastic beats. Refer to the beat release note for information.
The highlights in version 6.x include support for monitoring Kubernetes, new Metricbeat modules, and a better upgrade experience.
Metricbeat supports a new Kubernetes module that provides details on container pods such as CPU and memory usage, and network data, while Filebeat now includes a new processor that adds pod names, namespaces, and labels to shipped data.
New Metricbeat modules include memcached, perfmon, Elasticsearch, Kibana, vSphere and RabbitMQ.
- New Colors to Improve Accessibility
- Full Screen Mode for Dashboard
- Experimental Kibana Query Language (Kuery)
- Refactor of the Visualization Code
- new Application monitoring plugin
Punch Siddhi Integration
Siddhi rules can be invoked directly from punchlets. This makes it dramatically simpler to design complex event processing applications. Refer to the punch documentation.
We have performed an important revamp of our administration stack. The platform will rely on Elasticsearch and Kibana dashboards and plugins to provide users with the monitoring features :
- cluster and resource management
- health platform status and exposure to third party REST supervisors
- central log management of all platform services
- per tenant management of all administrative tasks
The new Craig architecture will be soon described in a dedicated blog post.
Avro dynamic record serialization/deserialisation and dynamic type advertisment will be available. This will permit significant gains in disk space disk or throughput, and a better compatibility between the punchplatform and third party components Avro enabled.
Moving to Avro format will require no coding action from users and will be activated only by configuration choice.
Punchplatform Spark and Machine Learning Sdk
We have been working long of this features that brings the power of the ES-Hadoop together with the Punchplatform Spark by configuration sdk. In addition the punchplatform will integrate out of the box two algorithm from Thales research labs, including the copulas.
This feature is called PunchPlatform Machine Learning (PML). Refer to the existing (beta) documentation.
The Kibana change will impact existing dashboards. Please refer to the kibana breaking changes page.
Elasticsearch upgrade from a Brad release will pose no particular operational issue. It is worth noticing that starting with Craig all Elasticsearch accesses will be implemented on top of the Rest api. The transport protocol will be unsupported. Note also that the index type field has been deprecated by Elastic. It is advised to adapt mappings and parsers should they use it, in a way to prepare the complete type index field depecation in Elastic future versions.
Get in touch with us firstname.lastname@example.org for preparing your migration.