• Tiered storage uses to support Amazon S3 and (GCS for short) for long term storage. With Jclouds, it is easy to add support for more cloud storage providers in the future.

  • Tiered storage uses to support filesystem for long term storage. With Hadoop, it is easy to add support for more filesystem in the future.

Tiered storage should be used when you have a topic for which you want to keep a very long backlog for a long time. For example, if you have a topic containing user actions which you use to train your recommendation systems, you may want to keep that data for a long time, so that if you change your recommendation algorithm you can rerun it against your full user history.

A topic in Pulsar is backed by a log, known as a managed ledger. This log is composed of an ordered list of segments. Pulsar only every writes to the final segment of the log. All previous segments are sealed. The data within the segment is immutable. This is known as a segment oriented architecture.

The Tiered Storage offloading mechanism takes advantage of this segment oriented architecture. When offloading is requested, the segments of the log are copied, one-by-one, to tiered storage. All segments of the log, apart from the segment currently being written to can be offloaded.

On the broker, the administrator must configure the bucket and credentials for the cloud storage service. The configured bucket must exist before attempting to offload. If it does not exist, the offload operation will fail.

Pulsar uses multi-part objects to upload the segment data. It is possible that a broker could crash while uploading the data. We recommend you add a life cycle rule your bucket to expire incomplete multi-part upload after a day or two to avoid getting charged for incomplete uploads.

Offloading is configured in .

At a minimum, the administrator must configure the driver, the bucket and the authenticating credentials. There is also some other knobs to configure, like the bucket region, the max block size in backed storage, etc.

Currently we support driver of types:

Bucket and Region

Buckets are the basic containers that hold your data. Everything that you store in Cloud Storage must be contained in a bucket. You can use buckets to organize your data and control access to your data, but unlike directories and folders, you cannot nest buckets.

  1. s3ManagedLedgerOffloadBucket=pulsar-topic-offload

Bucket Region is the region where bucket located. Bucket Region is not a required but a recommended configuration. If it is not configured, It will use the default region.

  1. s3ManagedLedgerOffloadRegion=eu-west-3

Authentication with AWS

To be able to access AWS S3, you need to authenticate with AWS S3. Pulsar does not provide any direct means of configuring authentication for AWS S3, but relies on the mechanisms supported by the DefaultAWSCredentialsProviderChain.

Once you have created a set of credentials in the AWS IAM console, they can be configured in a number of ways.

  1. Using ec2 instance metadata credentials

If you are on AWS instance with an instance profile that provides credentials, Pulsar will use these credentials if no other mechanism is provided

  1. Set the environment variables AWS_ACCESS_KEY_ID and AWS_SECRET_ACCESS_KEY in conf/pulsar_env.sh.
  1. export AWS_ACCESS_KEY_ID=ABC123456789
  2. export AWS_SECRET_ACCESS_KEY=ded7db27a4558e2ea8bbf0bf37ae0e8521618f366c
  1. Add the Java system properties aws.accessKeyId and aws.secretKey to PULSAR_EXTRA_OPTS in conf/pulsar_env.sh.
  1. PULSAR_EXTRA_OPTS="${PULSAR_EXTRA_OPTS} ${PULSAR_MEM} ${PULSAR_GC} -Daws.accessKeyId=ABC123456789 -Daws.secretKey=ded7db27a4558e2ea8bbf0bf37ae0e8521618f366c -Dio.netty.leakDetectionLevel=disabled -Dio.netty.recycler.maxCapacity.default=1000 -Dio.netty.recycler.linkCapacity=1024"
  1. Set the access credentials in ~/.aws/credentials.
  2. aws_access_key_id=ABC123456789
  3. aws_secret_access_key=ded7db27a4558e2ea8bbf0bf37ae0e8521618f366c
  1. Assuming an IAM role

If you want to assume an IAM role, this can be done via specifying the following:

This will use the DefaultAWSCredentialsProviderChain for assuming this role.

Configuring the size of block read/write

Pulsar also provides some knobs to configure the size of requests sent to AWS S3.

  • s3ManagedLedgerOffloadMaxBlockSizeInBytes configures the maximum size of a “part” sent during a multipart upload. This cannot be smaller than 5MB. Default is 64MB.
  • s3ManagedLedgerOffloadReadBufferSizeInBytes configures the block size for each individual read when reading back data from AWS S3. Default is 1MB.

In both cases, these should not be touched unless you know what you are doing.

Buckets are the basic containers that hold your data. Everything that you store in Cloud Storage must be contained in a bucket. You can use buckets to organize your data and control access to your data, but unlike directories and folders, you cannot nest buckets.

  1. gcsManagedLedgerOffloadBucket=pulsar-topic-offload

Bucket Region is the region where bucket located. Bucket Region is not a required but a recommended configuration. If it is not configured, It will use the default region.

Regarding GCS, buckets are default created in the us multi-regional location, page Bucket Locations contains more information.

  1. gcsManagedLedgerOffloadRegion=europe-west3

Authentication with GCS

The administrator needs to configure gcsManagedLedgerOffloadServiceAccountKeyFile in broker.conf for the broker to be able to access the GCS service. gcsManagedLedgerOffloadServiceAccountKeyFile is a Json file, containing the GCS credentials of a service account. Service Accounts section of this page contains more information of how to create this key file for authentication. More information about google cloud IAM is available .

To generate service account credentials or view the public credentials that you’ve already generated, follow the following steps:

  1. Open the Service accounts page.
  2. Click Create service account.
  3. In the Create service account window, type a name for the service account, and select Furnish a new private key. If you want to to the service account, also select Enable G Suite Domain-wide Delegation.
  4. Click Create.
  1. gcsManagedLedgerOffloadServiceAccountKeyFile="/Users/hello/Downloads/project-804d5e6a6f33.json"

Configuring the size of block read/write

Pulsar also provides some knobs to configure the size of requests sent to GCS.

  • gcsManagedLedgerOffloadMaxBlockSizeInBytes configures the maximum size of a “part” sent during a multipart upload. This cannot be smaller than 5MB. Default is 64MB.
  • gcsManagedLedgerOffloadReadBufferSizeInBytes configures the block size for each individual read when reading back data from GCS. Default is 1MB.

In both cases, these should not be touched unless you know what you are doing.

Configure connection address

You can configure the connection address in the broker.conf file.

Configure Hadoop profile path

The configuration file is stored in the Hadoop profile path. It contains various settings, such as base path, authentication, and so on.

  1. fileSystemProfilePath="../conf/filesystem_offload_core_site.xml"

The model for storing topic data uses org.apache.hadoop.io.MapFile. You can use all of the configurations in org.apache.hadoop.io.MapFile for Hadoop.

示例

For more information about the configurations in org.apache.hadoop.io.MapFile, see .

Namespace policies can be configured to offload data automatically once a threshold is reached. The threshold is based on the size of data that the topic has stored on the pulsar cluster. Once the topic reaches the threshold, an offload operation will be triggered. Setting a negative value to the threshold will disable automatic offloading. Setting the threshold to 0 will cause the broker to offload data as soon as it possiby can.

  1. $ bin/pulsar-admin namespaces set-offload-threshold --size 10M my-tenant/my-namespace

Offloading can manually triggered through a REST endpoint on the Pulsar broker. We provide a CLI which will call this rest endpoint for you.

When triggering offload, you must specify the maximum size, in bytes, of backlog which will be retained locally on the bookkeeper. The offload mechanism will offload segments from the start of the topic backlog until this condition is met.

  1. $ bin/pulsar-admin topics offload --size-threshold 10M my-tenant/my-namespace/topic1
  2. Offload triggered for persistent://my-tenant/my-namespace/topic1 for messages before 2:0:-1

The command to triggers an offload will not wait until the offload operation has completed. To check the status of the offload, use offload-status.

  1. $ bin/pulsar-admin topics offload-status my-tenant/my-namespace/topic1
  2. Offload is currently running

To wait for offload to complete, add the -w flag.

  1. $ bin/pulsar-admin topics offload-status -w my-tenant/my-namespace/topic1
  2. Offload was a success

If there is an error offloading, the error will be propagated to the offload-status command.

  1. $ bin/pulsar-admin topics offload-status persistent://public/default/topic1
  2. Error in offload
  3. null