- Introducing Elasticsearch Service
- Adding data to Elasticsearch
- Migrating data
- Ingesting data from your application
- Ingest data with Node.js on Elasticsearch Service
- Ingest data with Python on Elasticsearch Service
- Ingest data from Beats to Elasticsearch Service with Logstash as a proxy
- Ingest data from a relational database into Elasticsearch Service
- Ingest logs from a Python application using Filebeat
- Ingest logs from a Node.js web application using Filebeat
- Configure Beats and Logstash with Cloud ID
- Best practices for managing your data
- Configure index management
- Enable cross-cluster search and cross-cluster replication
- Access other deployments of the same Elasticsearch Service organization
- Access deployments of another Elasticsearch Service organization
- Access deployments of an Elastic Cloud Enterprise environment
- Access clusters of a self-managed environment
- Enabling CCS/R between Elasticsearch Service and ECK
- Edit or remove a trusted environment
- Migrate the cross-cluster search deployment template
- Manage data from the command line
- Preparing a deployment for production
- Securing your deployment
- Monitoring your deployment
- Monitor with AutoOps
- Configure Stack monitoring alerts
- Access performance metrics
- Keep track of deployment activity
- Diagnose and resolve issues
- Diagnose unavailable nodes
- Why are my shards unavailable?
- Why is performance degrading over time?
- Is my cluster really highly available?
- How does high memory pressure affect performance?
- Why are my cluster response times suddenly so much worse?
- How do I resolve deployment health warnings?
- How do I resolve node bootlooping?
- Why did my node move to a different host?
- Snapshot and restore
- Managing your organization
- Your account and billing
- Billing Dimensions
- Billing models
- Using Elastic Consumption Units for billing
- Edit user account settings
- Monitor and analyze your account usage
- Check your subscription overview
- Add your billing details
- Choose a subscription level
- Check your billing history
- Update billing and operational contacts
- Stop charges for a deployment
- Billing FAQ
- Elasticsearch Service hardware
- Elasticsearch Service GCP instance configurations
- Elasticsearch Service GCP default provider instance configurations
- Elasticsearch Service AWS instance configurations
- Elasticsearch Service AWS default provider instance configurations
- Elasticsearch Service Azure instance configurations
- Elasticsearch Service Azure default provider instance configurations
- Change hardware for a specific resource
- Elasticsearch Service regions
- About Elasticsearch Service
- RESTful API
- Release notes
- Enhancements and bug fixes - December 2024
- Enhancements and bug fixes - November 2024
- Enhancements and bug fixes - Late October 2024
- Enhancements and bug fixes - Early October 2024
- Enhancements and bug fixes - September 2024
- Enhancements and bug fixes - Late August 2024
- Enhancements and bug fixes - Early August 2024
- Enhancements and bug fixes - July 2024
- Enhancements and bug fixes - Late June 2024
- Enhancements and bug fixes - Early June 2024
- Enhancements and bug fixes - Early May 2024
- Bring your own key, and more
- AWS region EU Central 2 (Zurich) now available
- GCP region Middle East West 1 (Tel Aviv) now available
- Enhancements and bug fixes - March 2024
- Enhancements and bug fixes - January 2024
- Enhancements and bug fixes
- Enhancements and bug fixes
- Enhancements and bug fixes
- Enhancements and bug fixes
- AWS region EU North 1 (Stockholm) now available
- GCP regions Asia Southeast 2 (Indonesia) and Europe West 9 (Paris)
- Enhancements and bug fixes
- Enhancements and bug fixes
- Bug fixes
- Enhancements and bug fixes
- Role-based access control, and more
- Newly released deployment templates for Integrations Server, Master, and Coordinating
- Enhancements and bug fixes
- Enhancements and bug fixes
- Enhancements and bug fixes
- Enhancements and bug fixes
- Enhancements and bug fixes
- Enhancements and bug fixes
- Enhancements and bug fixes
- Enhancements and bug fixes
- Enhancements and bug fixes
- Enhancements and bug fixes
- Cross environment search and replication, and more
- Enhancements and bug fixes
- Enhancements and bug fixes
- Azure region Canada Central (Toronto) now available
- Azure region Brazil South (São Paulo) now available
- Azure region South Africa North (Johannesburg) now available
- Azure region Central India (Pune) now available
- Enhancements and bug fixes
- Azure new virtual machine types available
- Billing Costs Analysis API, and more
- Organization and billing API updates, and more
- Integrations Server, and more
- Trust across organizations, and more
- Organizations, and more
- Elastic Consumption Units, and more
- AWS region Africa (Cape Town) available
- AWS region Europe (Milan) available
- AWS region Middle East (Bahrain) available
- Enhancements and bug fixes
- Enhancements and bug fixes
- GCP Private Link, and more
- Enhancements and bug fixes
- GCP region Asia Northeast 3 (Seoul) available
- Enhancements and bug fixes
- Enhancements and bug fixes
- Native Azure integration, and more
- Frozen data tier and more
- Enhancements and bug fixes
- Azure region Southcentral US (Texas) available
- Azure region East US (Virginia) available
- Custom endpoint aliases, and more
- Autoscaling, and more
- Cross-region and cross-provider support, warm and cold data tiers, and more
- Better feature usage tracking, new cost and usage analysis page, and more
- New features, enhancements, and bug fixes
- AWS region Asia Pacific (Hong Kong)
- Enterprise subscription self service, log in with Microsoft, bug fixes, and more
- SSO for Enterprise Search, support for more settings
- Azure region Australia East (New South Wales)
- New logging features, better GCP marketplace self service
- Azure region US Central (Iowa)
- AWS region Asia Pacific (Mumbai)
- Elastic solutions and Microsoft Azure Marketplace integration
- AWS region Pacific (Seoul)
- AWS region EU West 3 (Paris)
- Traffic management and improved network security
- AWS region Canada (Central)
- Enterprise Search
- New security setting, in-place configuration changes, new hardware support, and signup with Google
- Azure region France Central (Paris)
- Regions AWS US East 2 (Ohio) and Azure North Europe (Ireland)
- Our Elasticsearch Service API is generally available
- GCP regions Asia East 1 (Taiwan), Europe North 1 (Finland), and Europe West 4 (Netherlands)
- Azure region UK South (London)
- GCP region US East 1 (South Carolina)
- GCP regions Asia Southeast 1 (Singapore) and South America East 1 (Sao Paulo)
- Snapshot lifecycle management, index lifecycle management migration, and more
- Azure region Japan East (Tokyo)
- App Search
- GCP region Asia Pacific South 1 (Mumbai)
- GCP region North America Northeast 1 (Montreal)
- New Elastic Cloud home page and other improvements
- Azure regions US West 2 (Washington) and Southeast Asia (Singapore)
- GCP regions US East 4 (N. Virginia) and Europe West 2 (London)
- Better plugin and bundle support, improved pricing calculator, bug fixes, and more
- GCP region Asia Pacific Southeast 1 (Sydney)
- Elasticsearch Service on Microsoft Azure
- Cross-cluster search, OIDC and Kerberos authentication
- AWS region EU (London)
- GCP region Asia Pacific Northeast 1 (Tokyo)
- Usability improvements and Kibana bug fix
- GCS support and private subscription
- Elastic Stack 6.8 and 7.1
- ILM and hot-warm architecture
- Elasticsearch keystore and more
- Trial capacity and more
- APM Servers and more
- Snapshot retention period and more
- Improvements and snapshot intervals
- SAML and multi-factor authentication
- Next generation of Elasticsearch Service
- Branding update
- Minor Console updates
- New Cloud Console and bug fixes
- What’s new with the Elastic Stack
Deployment autoscaling
editDeployment autoscaling
editAutoscaling helps you to more easily manage your deployments by adjusting their available resources automatically, and currently supports scaling for both data and machine learning nodes, or machine learning nodes only. Check the following sections to learn more:
You can also have a look at our autoscaling example, as well as a sample request to create an autoscaled deployment through the API.
Overview
editWhen you first create a deployment it can be challenging to determine the amount of storage your data nodes will require. The same is relevant for the amount of memory and CPU that you want to allocate to your machine learning nodes. It can become even more challenging to predict these requirements for weeks or months into the future. In an ideal scenario, these resources should be sized to both ensure efficient performance and resiliency, and to avoid excess costs. Autoscaling can help with this balance by adjusting the resources available to a deployment automatically as loads change over time, reducing the need for monitoring and manual intervention.
Autoscaling is enabled for the Machine Learning tier by default for new deployments.
Currently, autoscaling behavior is as follows:
-
Data tiers
- Each Elasticsearch data tier scales upward based on the amount of available storage. When we detect more storage is needed, autoscaling will scale up each data tier independently to ensure you can continue and ingest more data to your hot and content tier, or move data to the warm, cold, or frozen data tiers.
- In addition to scaling up existing data tiers, a new data tier will be automatically added when necessary, based on your index lifecycle management policies.
- To control the maximum size of each data tier and ensure it will not scale above a certain size, you can use the maximum size per zone field.
- Autoscaling based on memory or CPU, as well as autoscaling downward, is not currently supported. In case you want to adjust the size of your data tier to add more memory or CPU, or in case you deleted data and want to scale it down, you can set the current size per zone of each data tier manually.
-
Machine learning nodes
- Machine learning nodes can scale upward and downward based on the configured machine learning jobs.
- When a machine learning job is opened, or a machine learning trained model is deployed, if there are no machine learning nodes in your deployment, the autoscaling mechanism will automatically add machine learning nodes. Similarly, after a period of no active machine learning jobs, any enabled machine learning nodes are disabled automatically.
- To control the maximum size of your machine learning nodes and ensure they will not scale above a certain size, you can use the maximum size per zone field.
- To control the minimum size of your machine learning nodes and ensure the autoscaling mechanism will not scale machine learning below a certain size, you can use the minimum size per zone field.
- The determination of when to scale is based on the expected memory and CPU requirements for the currently configured machine learning jobs and trained models.
For any Elasticsearch Service Elasticsearch component the number of availability zones is not affected by autoscaling. You can always set the number of availability zones manually and the autoscaling mechanism will add or remove capacity per availability zone.
When does autoscaling occur?
editSeveral factors determine when data tiers or machine learning nodes are scaled.
For a data tier, an autoscaling event can be triggered in the following cases:
- Based on an assessment of how shards are currently allocated, and the amount of storage and buffer space currently available.
When past behavior on a hot tier indicates that the influx of data can increase significantly in the near future. Refer to Reactive storage decider and Proactive storage decider for more detail.
- Through ILM policies. For example, if a deployment has only hot nodes and autoscaling is enabled, it automatically creates warm or cold nodes, if an ILM policy is trying to move data from hot to warm or cold nodes.
On machine learning nodes, scaling is determined by an estimate of the memory and CPU requirements for the currently configured jobs and trained models. When a new machine learning job tries to start, it looks for a node with adequate native memory and CPU capacity. If one cannot be found, it stays in an opening
state. If this waiting job exceeds the queueing limit set in the machine learning decider, a scale up is requested. Conversely, as machine learning jobs run, their memory and CPU usage might decrease or other running jobs might finish or close. In this case, if the duration of decreased resource usage exceeds the set value for down_scale_delay
, a scale down is requested. Check Machine learning decider for more detail. To learn more about machine learning jobs in general, check Create anomaly detection jobs.
On a highly available deployment, autoscaling events are always applied to instances in each availability zone simultaneously, to ensure consistency.
Notifications
editIn the event that a data tier or machine learning node scales up to its maximum possible size, you’ll receive an email, and a notice also appears on the deployment overview page prompting you to adjust your autoscaling settings to ensure optimal performance.
Restrictions and limitations
editThe following are known limitations and restrictions with autoscaling:
- Autoscaling will not run if the cluster is unhealthy or if the last Elasticsearch plan failed.
- Trial deployments cannot be configured to autoscale beyond the normal Trial deployment size limits. The maximum size per zone is increased automatically from the Trial limit when you convert to a paid subscription.
- If Enterprise Search is left at the default size, and has no engines or sources configured for an extended period, it will be automatically disabled. This occurs whether autoscaling is enabled or not. Reactivate the service by editing the Enterprise Search settings for your deployment under Cloud > Deployments > your-deployment > Edit.
- ELSER deployments do not scale automatically. For more information, refer to ELSER and Trained model autoscaling.
Enable or disable autoscaling
editTo enable or disable autoscaling on a deployment:
- Log in to the Elasticsearch Service Console.
-
On the Deployments page, select your deployment.
On the deployments page you can narrow your deployments by name, ID, or choose from several other filters. To customize your view, use a combination of filters, or change the format from a grid to a list.
- In your deployment menu, select Edit.
- Select desired autoscaling configuration for this deployment using Enable Autoscaling for: dropdown menu.
- Select Confirm to have the autoscaling change and any other settings take effect. All plan changes are shown on the Deployment Activity page.
When autoscaling has been enabled, the autoscaled nodes resize according to the autoscaling settings. Current sizes are shown on the deployment overview page.
When autoscaling has been disabled, you need to adjust the size of data tiers and machine learning nodes manually.
Update your autoscaling settings
editEach autoscaling setting is configured with a default value. You can adjust these if necessary, as follows:
- Log in to the Elasticsearch Service Console.
-
On the Deployments page, select your deployment.
On the deployments page you can narrow your deployments by name, ID, or choose from several other filters. To customize your view, use a combination of filters, or change the format from a grid to a list.
- In your deployment menu, select Edit.
-
To update a data tier:
- Use the dropdown box to set the Maximum size per zone to the largest amount of resources that should be allocated to the data tier automatically. The resources will not scale above this value.
- You can also update the Current size per zone. If you update this setting to match the Maximum size per zone, the data tier will remain fixed at that size.
- For a hot data tier you can also adjust the Forecast window. This is the duration of time, up to the present, for which past storage usage is assessed in order to predict when additional storage is needed.
- Select Save to apply the changes to your deployment.
-
To update machine learning nodes:
- Use the dropdown box to set the Minimum size per zone and Maximum size per zone to the smallest and largest amount of resources, respectively, that should be allocated to the nodes automatically. The resources allocated to machine learning will not exceed these values. If you set these two settings to the same value, the machine learning node will remain fixed at that size.
- Select Save to apply the changes to your deployment.
You can also view our example of how the autoscaling settings work.
On this page
ElasticON events are back!
Learn about the Elastic Search AI Platform from the experts at our live events.
Register now