Storage
Applicable to open source & enterprise version
There are 2 primary items that need to be stored in OpenObserve.
- Ingested stream data
- Metadata of ingested stream data
Metadata for stream data is always stored on disk for Local mode and will stored in etcd for Cluster mode.
Stream data can be stored on disk, s3, minIO and other compatible s3 API object storages, like: Google GCS, Alibaba OSS, Tencent COS.
Tips:
- Default OpenObserve runs as
Local mode, you can setLOCAL_MODE=falseto enableCluster mode. - In
Local modeit also can uses3as storage, you can setZO_LOCAL_MODE_STORAGE=s3to storage data in s3. - For GCS, OSS they all supports
s3SDK, so you can think they are all of s3, just configure different s3 environments.
Data
Data is stored in parquet format, which is columnar storage format, it is efficient for query and storage.
Disk
Disk is default storage place for stream data, make sure you have enough storage available on your disk for stream data storage. During data ingestion the stream name that you provide may not be present, in that case a new folder for stream will be created automatically for storing the stream data.
S3
To use S3 for storing stream data following needs to be done:
-
Make AWS IAM credentials available to OpenObserve by any supported mechanisms. OpenObserve uses AWS SDK which looks for credentials through:
-
Environment variables AWS_ACCESS_KEY_ID and AWS_SECRET_ACCESS_KEY (Not recommended due to security concerns)
- Your aws CLI credentials stored in ~/.aws/credentials
- Instance metadata on an EC2 Instance/Fargate container/ECS container. You do this providing an IAM role to an EC2 instance or a task role to an ECS task - IMDS/IMDSv2 . ECS is not recommended as it does not have support for stateful workloads off EBS as of now.
- IAM Roles for service Accounts in EKS
You need to create the bucket in S3 first.
MinIO
OpenObserve can use MinIO for storing stream data, following environment variables needs to be setup:
| Environment Variable | Value | Description |
|---|---|---|
| ZO_S3_SERVER_URL | - | minIO server address |
| ZO_S3_REGION_NAME | - | region name, can be anything, like: us-west-1 |
| ZO_S3_ACCESS_KEY | - | access key |
| ZO_S3_SECRET_KEY | - | secret key |
| ZO_S3_BUCKET_NAME | - | bucket name |
| ZO_S3_PROVIDER | minio | used to specify settings like force_style=true |
You need to create the bucket in MinIO first.
Openstack Swift
OpenObserve can use Openstack swift for storing stream data, following environment variables needs to be setup:
| Environment Variable | Value | Description |
|---|---|---|
| ZO_S3_SERVER_URL | - | swift server address, like: https://us-west-1.xxx.com |
| ZO_S3_REGION_NAME | - | region name, can be anything, like: us-west-1 |
| ZO_S3_ACCESS_KEY | - | access key |
| ZO_S3_SECRET_KEY | - | secret key |
| ZO_S3_BUCKET_NAME | - | bucket name |
| ZO_S3_FEATURE_HTTP1_ONLY | true | -- |
| ZO_S3_PROVIDER | s3 | Use s3 compatible API |
| AWS_EC2_METADATA_DISABLED | true | swift doesn't support ec2 metadata, Usually you don't need this |
You need to create the bucket in swift first.
Google GCS
OpenObserve can use google cloud storage for storing stream data, following environment variables needs to be setup:
| Environment Variable | Value | Description |
|---|---|---|
| ZO_S3_SERVER_URL | - | gcs server address. should be: https://storage.googleapis.com |
| ZO_S3_REGION_NAME | - | region name, gcs region name, or: auto |
| ZO_S3_ACCESS_KEY | - | access key |
| ZO_S3_SECRET_KEY | - | secret key |
| ZO_S3_BUCKET_NAME | - | bucket name |
| ZO_S3_FEATURE_HTTP1_ONLY | true | -- |
| ZO_S3_PROVIDER | s3 | Use s3 compatible API for gcp |
You can refer to: https://cloud.google.com/storage/docs/aws-simple-migration
Alibaba OSS (aliyun)
OpenObserve can use Alibaba(aliyun) OSS for storing stream data, following environment variables needs to be setup:
| Environment Variable | Value | Description |
|---|---|---|
| ZO_S3_SERVER_URL | - | oss endpoint address, eg: https://oss-cn-beijing.aliyuncs.com |
| ZO_S3_REGION_NAME | - | region name, oss region name, eg: oss-cn-beijing |
| ZO_S3_ACCESS_KEY | - | access key |
| ZO_S3_SECRET_KEY | - | secret key |
| ZO_S3_BUCKET_NAME | - | bucket name |
| ZO_S3_FEATURE_FORCE_PATH_STYLE | true | -- |
You can refer to: https://help.aliyun.com/document_detail/64919.html
Tencent COS
OpenObserve can use tencent cloud storage for storing stream data, following environment variables needs to be setup:
| Environment Variable | Value | Description |
|---|---|---|
| ZO_S3_SERVER_URL | - | cos endpoint address |
| ZO_S3_REGION_NAME | - | region name, cos region name |
| ZO_S3_ACCESS_KEY | - | access key |
| ZO_S3_SECRET_KEY | - | secret key |
| ZO_S3_BUCKET_NAME | - | bucket name |
You can refer to: https://cloud.tencent.com/document/product/436/37421
Baidu BOS
OpenObserve can use baidu cloud storage for storing stream data, following environment variables needs to be setup:
| Environment Variable | Value | Description |
|---|---|---|
| ZO_S3_SERVER_URL | - | bos endpoint address, eg: https://s3.bj.bcebos.com |
| ZO_S3_REGION_NAME | - | region name, bos region name, eg: bj |
| ZO_S3_ACCESS_KEY | - | access key |
| ZO_S3_SECRET_KEY | - | secret key |
| ZO_S3_BUCKET_NAME | - | bucket name |
You can refer to: https://cloud.baidu.com/doc/BOS/s/xjwvyq9l4
Azure Blob
OpenObserve can use azure blob for storing stream data. Following environment variables needs to be setup:
| Environment Variable | Value | Description |
|---|---|---|
| ZO_S3_PROVIDER | azure | Azure blob storage provider |
| ZO_LOCAL_MODE_STORAGE | s3 | Required only if running in single node mode |
| AZURE_STORAGE_ACCOUNT_NAME | Storage account name | Need to provide mandatorily |
| AZURE_STORAGE_ACCESS_KEY | Access key | Need to provide mandatorily |
| ZO_S3_BUCKET_NAME | Blob Container name | Need to provide mandatorily |
Metadata
Default meta store is SQLite and can be changed by using ZO_META_STORE environment variable.
etcd
ZO_META_STORE=etcd
While etcd is used as the cluster coordinator it is also set as the default meta store by the older helm charts. Newer helm charts released after Feb 23 2024 use postgres as the meta store.
SQLite
ZO_META_STORE=sqlite
Installations running on a single node can use SQLite as metadata store. You do not need to take any extra steps to use SQLite. This is generally not recommended as losing the SQLite data will causing OpenObserve to not operate.
PostgreSQL
ZO_META_STORE=postgres
This is the recommended meta store as it is more reliable and scalable. A lot of service providers provide managed postgres service that you can rely upon.
Default helm chart released after Feb 23, 2024 use cloudnative-pg to create a postgres cluster (primary + replica) which is used as the meta store. these instances provide great HA capability and backup and restore is much easier too in case its required.
MySQL
ZO_META_STORE=mysql
Use this if you prefer MySQL for any reason.