GET hardware requirements
Enterprise

GET hardware requirements

This page outlines the current hardware requirements for running Grafana Enterprise Traces (GET). Grafana Labs reserves the right to mark a support issue as ‘unresolvable’ if these requirements are not followed. Refer to the Grafana Labs Enterprise Support SLA for more details.

Note

As of February 2025, you must run GET on the managed Kubernetes services offered by AWS (Amazon Elastic Kubernetes Service), GCP (Google Kubernetes Engine), or Azure (Azure Kubernetes Service). Additionally, you must use the managed object storage services offered by AWS (S3), GCP (GCS), or Azure (Blob Storage).

No other deployment modes are supported.

CPU and memory

GET should be deployed on machines with a 1:4 ratio of CPU to memory. For every CPU core, you should have 4GB of memory.

Grafana Labs recommends deploying GET onto machines with at least 16 CPU cores and 64GB of memory. All the nodes in the Kubernetes cluster running GET should be of the same type.

Network

All components of GET require fast network access. Nodes the software runs on should be connected by 10 gigabit/second or faster network connection speed.

Storage

GET requires fast storage to run. Any storage solutions you use have to have adequate performance. Object storage (for example AWS S3 buckets, Google Cloud GCS or Microsoft Azure Storage) buckets or blobs are used to persist trace data for the required retention period. Your object storage should be large enough to contain 30 days worth of data, the default block retention for GET.

However, various components of GET (for example, the ingester) also require fast, persistent block storage to be available to the host machine. For example, in the case of the ingester component, all incoming data is sent to a write-ahead log (WAL) to aggregate trace data into blocks before storing them in object storage, and to help withstand unexpected node termination.

For every deployment, you need both object and block storage. Block storage must be memory-state storage, like SSD or NVMEs. Hard drive (platter-based) disk storage is not supported.

The supported configurations for AWS, GCP, and Azure are in the following sections.

Block storage

PlatformStorage typeTested withComments
Amazon Web ServicesBlockio1 provisioned IOPS SSD EBS volumesThe io1 storage must be provisioned at 50 IOPS per gigabyte, with a minimum of 150Gi allocated to ensure performant I/O.
Google Cloud PlatformBlockpd-ssd SSD persistent storage
Microsoft AzureBlockPremium SSD SSD persistent storage

Object storage

GET requires object storage for configuration storage, as well as long-term data storage. Amazon AWS S3, GCP GCS, and Azure Blob storage are the only supported object stores for running GET.

PlatformStorage typeTested withComments
Amazon Web ServicesS3S3 StandardS3 object storage service using the Standard storage class.
Google Cloud PlatformGCSSTANDARDGCS object storage service using the STANDARD storage class in both regional and dual regional storage locations.
Microsoft AzureAzure BlobHot or Cool tier storage class with LRS replicationBlob Storage object storage service using the Hot or Cool tier with replication type Locally Redundant Storage (LRS)