Elastic Cloud Enterprise (ECE) has hit 2.0, with a focus on facilitating logging and working with time-series data, so that ops have an easier time in complex setups.
The updated version also lets users handle cluster nodes separately and offers instance configuration, which can be used to set the roles of a node or find allocator hosts.
To realise the latter, Elasticsearch has added allocator tags in v2.0. Tags can include information on the underlying hardware or the team it belongs to, and are used to match allocators not only to nodes but Kibana instances for hosting them. Instance configurations can also be grouped into deployment templates, which let admins define defaults like node size for the different node types and use cases.
With the introduction of instance configuration, there are now also ways to run machine learning jobs on dedicated, independently scalable nodes, which wasn’t an option before. It also gives an option to create master nodes to handle large clusters, or special nodes for data consumption.
Architecture templates and usability improvements
Since the team behind the Elastic Stack recommends a so-called hot-warm architecture for time series data, ECE now offers a template for such use cases. The architecture consists of nodes that ingest data and store recent indices (hot data nodes) and those for longer term data retention (warm data nodes). Since data with more recent indices is assumed to be queried more often, it should be deployed on a stable host with a high I/O rate.
To make working with the Elastic Cloud easier, the navigation has been improved, and the Elasticsearch team has created a night theme. The newly added “create deployment” workflow should serve as a guide for configuring as well as creating new deployments, and save new defaults.
Since ECE 1.x only supported Docker 1.11, which has already reached its end of life, it’s important to note that the new version can be installed on Ubuntu 16.04 with Docker 18.03. More regular Docker updates are planned.
Elastic Cloud Enterprise is Elasticsearch’s on-premises version of the hosted Elastic Stack service. It can be used to provision, manage, and monitor search and analytics engine Elasticsearch as well as visualisation tool Kibana in different infrastructure setups – virtual environments, public or private clouds for example. Pricing is available on request only, though the first 14 days are free. Elasticsearch and Kibana themselves are open source projects whose elements are variously licensed under the Apache License Version 2.0 or the Elastic License.