How to Create a HarperDB Cluster

HarperDB is a versatile database solution that combines SQL and NoSQL functionality. It includes a comprehensive built-in API for easy integration with other applications. This guide provides a brief introduction to HarperDB and explains how to install it. It also explains how to configure multiple database instances into a cluster and replicate data.

What is HarperDB?

HarperDB combines a flexible database, built-in API, and distribution logic into a single backend. This solution, known as an embedded database, allows developers to more quickly and easily create integrated web applications. HarperDB allows both NoSQL and SQL tables to be mixed together in the same database and schema. SQL tables are highly structured and normalized, while NoSQL permits more freeform data. This combination enables access to legacy data and operational systems in the same place as new business intelligence analytics.

HarperDB is available through the HarperDB Cloud or as a self-hosted solution. The optional HarperDB Studio provides a visual GUI for storing or retrieving data but requires registration. Users can configure HarperDB through either the comprehensive API or the HarperDB CLI. Unfortunately, the CLI only supports a subset of the current functionality. API calls can be embedded into an application or sent as stand-alone requests using curl or a similar utility.

Note

Users can send most SQL commands to HarperDB using the API, though other query languages are preferred. Review HarperDB SQL Guide for more details.

HarperDB is optimized for fast performance and scalability, with sub-millisecond latency between the API and data layer. NoSQL data can be accessed as quickly as SQL tables in traditional relational database management systems (RDBMS). HarperDB is particularly useful for gaming, media, manufacturing, status reporting, and real-time updates.

HarperDB also supports clustering and per-table data replication. Data replication can be configured in one or both directions. Administrators have full control over how the data is replicated within the cluster. A HarperDB instance can both send table updates to a second node and receive updates from it in return. However, it can simultaneously transmit changes to a second table to another node in a unidirectional manner. HarperDB minimizes data latency between nodes, allowing clusters encompassing different regions and different continents. Clusters can grow very large, permitting virtually unlimited horizontal scaling.

Some advantages of HarperDB are:

The HarperDB API provides applications with direct database access. This allows the application and its data to be bundled together in a single distribution.
Each HarperDB node is atomic and guarantees “exactly-once” delivery. It avoids unnecessary data duplication.
Every node in the cluster can read, write, and replicate data.
HarperDB features a fast and resilient caching mechanism.
Connections are self-healing, allowing for fast replication even in an unstable network.
HarperDB supports data streaming and edge processing. This technique pre-processes data, only storing or transmitting the most important information.
NoSQL tables support dynamic schemas, which can seamlessly change as new data arrives. HarperDB provides an auto-indexing function for more efficient hashing.
HarperDB allows SQL queries on both structured and unstructured data.
HarperDB’s Custom Functions allow developers to add their own API endpoints and manage authentication and authorization.

Before You Begin

If you have not already done so, create a Linode account and Compute Instance. See our Getting Started with Linode and Creating a Compute Instance guides.
Follow our Setting Up and Securing a Compute Instance guide to update your system. You may also wish to set the timezone, configure your hostname, create a limited user account, and harden SSH access.
On a multi-user system, it is best to create a dedicated HarperDB user account with sudo access. Use this account for the instructions in this guide.

Note

The steps in this guide are written for non-root users. Commands that require elevated privileges are prefixed with sudo. If you are not familiar with the sudo command, see the Linux Users and Groups guide.

How To Install HarperDB

Run these instructions on every node in the cluster. Each cluster must contain at least two nodes. These guidelines are designed for Ubuntu 22.04 LTS users but are similar to other Linux distributions. HarperDB is also available as a Docker container or as a .tgz file for offline installation. For more details on these options and the standard installation procedure, see the HarperDB installation instructions .

Ensure the system is up to date by executing the following command:
sudo apt-get update -y && sudo apt-get upgrade -y
HarperDB requires Node.js to run properly. To install Node.js, first install the Node Version Manager (NVM). To download and install NVM, use the following command.
curl -o- https://raw.githubusercontent.com/nvm-sh/nvm/v0.39.3/install.sh | bash
Log out and log back into the terminal to activate NVM.
exit ssh username@system_IP
Use NVM to install Node.js. This command installs Node.js release 20, the current LTS release as of this writing. It also installs the NPM package manager for Node.js.
Note
HarperDB requires Node.js release 14 or higher.
nvm install 20
NVM installs the latest Node.js 20 LTS patch version and sets that as the default.

Create a swap file for the system.

sudo dd if=/dev/zero of=/swapfile bs=128M count=16
sudo chmod 600 /swapfile
sudo mkswap /swapfile
sudo swapon /swapfile
echo "/swapfile swap swap defaults 0 0" | sudo tee -a /etc/fstab

Increase the open file limits for the account. Replace accountname with the name of the actual account.

echo "accountname soft nofile 500000" | sudo tee -a /etc/security/limits.conf
echo "accountname hard nofile 1000000" | sudo tee -a /etc/security/limits.conf

Use NPM to install HarperDB.
npm install -g harperdb

How To Configure and Initialize the HarperDB Cluster

This section explains the steps required to initialize and run HarperDB. It also describes the additional configuration required to create and enable a HarperDB cluster. Each cluster must contain at least two nodes.

Some cluster attributes can be passed as parameters to the initial harperdb start command. If the system is initially configured as a stand-alone instance, it can be added to a cluster later on. However, further changes cannot be made through the command line. They must be implemented in either the harperdb-config.yaml file or through API calls.

For simplicity and consistency, this guide appends most of the required cluster configuration to the initial harperdb start command. It then completes the configuration process using API calls. For more information on the HarperDB API, see the HarperDB API documentation .

Note

Configuration tasks can also be accomplished through the HarperDB Studio GUI. HarperDB Studio requires a HarperDB user account and registration.

Replication occurs on a per-table basis and is configured after the schema and table are defined. See the following section for a more complete explanation. Follow these steps to enable clustering on your HarperDB nodes.

On the first node, use harperdb start to launch the application. Provide the following configuration attributes in the command.
- For TC_AGREEMENT, indicate yes to accept the terms of the agreement.
- Define the ROOTPATH directory for persistent data. This example sets the directory to /home/user/hdb. Replace the user with the actual name of the user account.
- Set the HDB_ADMIN_USERNAME to the name of the administrative user.
- Provide a password for the administrative account in HDB_ADMIN_PASSWORD. Replace the password with a more secure password.
- Set OPERATIONSAPI_NETWORK_PORT to 9925.
- Choose a name for the CLUSTERING_USER and provide a password for the user in CLUSTERING_PASSWORD. These values must be the same for all nodes in the cluster.
- Set CLUSTERING_ENABLED to true.
- Identify the node using CLUSTERING_NODENAME. This name must be unique within the cluster.
Note
HTTPS is recommended for better security on production systems or with sensitive data. To use HTTPS, add the parameters --OPERATIONSAPI_NETWORK_HTTPS "true" and --CUSTOMFUNCTIONS_NETWORK_HTTPS "true".
harperdb start \ --TC_AGREEMENT "yes" \ --ROOTPATH "/home/user/hdb" \ --OPERATIONSAPI_NETWORK_PORT "9925" \ --HDB_ADMIN_USERNAME "HDB_ADMIN" \ --HDB_ADMIN_PASSWORD "password" \ --CLUSTERING_ENABLED "true" \ --CLUSTERING_USER "cluster_user" \ --CLUSTERING_PASSWORD "password" \ --CLUSTERING_NODENAME "hdb1"
```
|------------- HarperDB 4.1.2 successfully started ------------|
```
(Optional) To launch HarperDB at bootup, create a crontab entry for the application. Substitute the name of the administrative account for user and ensure the path reflects the release of the NVM being used. In this example, the path entry reflects release 18.17 of NVM.
Note
To integrate HarperDB with systemd and start/stop it using systemctl, see the HarperDB Linux documentation .
(crontab -l 2>/dev/null; echo "@reboot PATH=\"/home/user/.nvm/versions/node/v18.17.0/bin:$PATH\" && harperdb start") | crontab -

Start HarperDB on the remaining nodes. Change the value of CLUSTERING_NODENAME to a different value. In this example, it is set to hdb2. The remaining attributes are the same as on the first node.

harperdb start \
--TC_AGREEMENT "yes" \
--ROOTPATH "/home/user/hdb" \
--OPERATIONSAPI_NETWORK_PORT "9925" \
--HDB_ADMIN_USERNAME "HDB_ADMIN" \
--HDB_ADMIN_PASSWORD "password" \
--CLUSTERING_ENABLED "true" \
--CLUSTERING_USER "cluster_user" \
--CLUSTERING_PASSWORD "password" \
--CLUSTERING_NODENAME "hdb2"

Run harperdb status on each node to confirm HarperDB is active. The status field should indicate running.

harperdb status

harperdb:
  status: running
  pid: 1726
clustering:
  hub server:
    status: running
    pid: 1698
  leaf server:
    status: running
    pid: 1715
  network:
    - name: hdb1
    response time: 6
    connected nodes: []
    routes: []
  replication:
    node name: hdb1
    is enabled: true
    connections: []

Determine the network topology for the cluster. A full mesh of connections is not required. Data can be replicated to any cluster node provided it is connected to the rest of the cluster. Design some measure of resiliency into the network. If a hub-and-spoke architecture is configured, the remaining nodes would be isolated if the central node suffers an outage. As a general guideline, connect each node to two other nodes. It is not necessary to add the route in both directions. For instance, a connection between node1 and node2 can be added to either node1 or node2. Successful negotiation establishes a bidirectional route.
Authentication is required to send messages to HarperDB using the API. To derive the AuthorizationKey from the name and password of the administrator account, use the JavaScript btoa() function. Run the command btoa("HDB_ADMIN:password") to convert the account credentials into a Base64 string. Replace the password with the actual password.
Note
JavaScript commands can be executed in a web browser console. On Firefox, select Tools->Browser Tools->Web Developer Tools to access the console. Choose the Console option within the developer window, then enter the command. Alternatively, online JavaScript emulators are widely available for the same purpose. Use the result for the AuthorizationKey values in the following API calls. See the Mozilla documentation for more information.
btoa("HDB_ADMIN:password")
Add routes until the network architecture is fully implemented. If a cluster consists of node, node2, and node3, add a route on node1 to reach node2 and another on node2 to node3. On node hdb1, run the curl command shown below to install a route to hdb2. Include the following information:
- In the POST header, send the command to the local HarperDB process at http://localhost:9925.
- Include an Authorization header. Use the AuthorizationKey derived from the administrator account and password in the previous step.
- Inside the data header, set the operation to cluster_set_routes and set the server to `hub``.
- Use routes to specify a list of one or more routes to install. Each route consists of a host and a port, which is typically 9932. The host is the IP address of the peer system. In the following example, replace 192.0.2.10 with the actual IP address of the peer.
curl --location --request POST 'http://localhost:9925' \ --header 'Authorization: Basic AuthorizationKey' \ --header 'Content-Type: application/json' \ --data '{ "operation": "cluster_set_routes", "server": "hub", "routes":[ {"host": "192.0.2.10", "port": 9932} ] }'
```
{"message":"cluster routes successfully set","set":[{"host":"192.0.2.10","port":9932}],"skipped":[]}
```
Stop and start the HarperDB instance to quickly negotiate the route.
harperdb stop harperdb start

Run the harperdb status command again. Ensure the route is displayed under routes.

harperdb status

harperdb:
  status: running
        pid: 20926
clustering:
  hub server:
    status: running
    pid: 20899
  leaf server:
    status: running
    pid: 20914
  network:
    - name: hdb1
    response time: 18
    connected nodes:
      - hdb2
    routes:
      - host: 192.0.2.10
        port: 9932
    - name: hdb2
    response time: 92
    connected nodes:
      - hdb1
    routes: []
  replication:
    node name: hdb1
    is enabled: true
    connections: []

How to Add and Replicate Data on HarperDB

The cluster is now ready for replication. Replication occurs on a per-table basis in HarperDB, so data is not automatically replicated. Instead, one or more subscriptions define how to manage the table data. The schema and table must be created first before adding any subscriptions. Each subscription references a single peer node. To replicate data to multiple nodes, multiple subscriptions must be added.

A subscription contains the name of the schema and table to replicate, along with Boolean values for publish and subscribe. When publish is set to true, transactions on the local node are replicated to the remote node. Setting subscribe to true means any changes to the remote table are sent to the local node. Both values can be set to true, resulting in bidirectional replication. In all cases, the local node is the one receiving the subscription request.

The following example demonstrates how to create a schema, table, and subscription on node hdb1. The subscription both publishes and subscribes to the dog table, resulting in two-way replication between nodes hdb1 and hdb2.

Create the dev schema on node hdb1 through the HarperDB API using the create_schema operation. Provide the correct value for the AuthorizationKey as described earlier.

curl --location --request POST 'http://localhost:9925' \
--header 'Authorization: Basic AuthorizationKey' \
--header 'Content-Type: application/json' \
--data '{
    "operation": "create_schema",
    "schema": "dev"
}'

{"message":"schema 'dev' successfully created"}

Create the dog table within the dev schema. This API call invokes the create_table operation and sets the hash_attribute to id. This is a NoSQL table, so columns and types are not defined.

curl --location --request POST 'http://localhost:9925' \
--header 'Authorization: Basic AuthorizationKey' \
--header 'Content-Type: application/json' \
--data '{
    "operation": "create_table",
    "schema": "dev",
    "table": "dog",
    "hash_attribute": "id"
}'

{"message":"table 'dev.dog' successfully created."}

Add a subscription to the dog table using the API add_node operation. Add the following information to the request.
- Set node_name to hdb2 to designate it as the peer for replication.
- Specify the schema and table to replicate. In this example, the schema is dev and the table is a dog.
- To transmit updates to hdb2 set publish to true. This configures replication in one direction only.
Note
The add_node operation can create multiple subscriptions for several schemas/tables at the same time. However, all subscriptions in the request must relate to the same peer. Separate each subscription using a comma and enclose it with the [] brackets. To replicate more tables to a different node, call the add_node API again and provide the new node_name.
curl --location --request POST 'http://localhost:9925' \ --header 'Authorization: Basic AuthorizationKey' \ --header 'Content-Type: application/json' \ --data '{ "operation": "add_node", "node_name": "hdb2", "subscriptions": [ { "schema": "dev", "table": "dog", "subscribe": false, "publish": true } ] }'
```
{"message":"Successfully added 'hdb2' to manifest","added":[{"schema":"dev","table":"dog","publish":true,"subscribe":false}],"skipped":[]}
```

Use harperdb status to confirm HarperDB is aware of the subscription.

harperdb status

...
  replication:
    node name: hdb1
    is enabled: true
    connections:
      - node name: hdb2
        status: open
        ports:
          clustering: 9932
          operations api: 9925
        latency ms: 132
        uptime: 6h 49m 43s
        subscriptions:
        - schema: dev
          table: dog
          publish: true
          subscribe: false

To subscribe to updates to the dog table from hdb2, use the update_node operation. Set both subscribe and publish to true in the API call.

Note

subscribe and publish could have been both set to true in the original add_node operation. This method demonstrates how to update an existing subscription. To completely remove the subscription, use the remove_node operation and include the name of the node under node_name.

curl --location --request POST 'http://localhost:9925' \
--header 'Authorization: Basic AuthorizationKey' \
--header 'Content-Type: application/json' \
--data '{
    "operation": "update_node",
    "node_name": "hdb2",
    "subscriptions": [
        {
            "schema": "dev",
            "table": "dog",
            "subscribe": true,
            "publish": true
        }
    ]
}'

{"message":"Successfully updated 'hdb2'","updated":[{"schema":"dev","table":"dog","publish":true,"subscribe":true}],"skipped":[]}

Add a record to the table to ensure replication is working. Either SQL or NoSQL can be used to add data to the dog table. This example adds a record using the NoSQL insert operation. Specify dev as the schema and dog as the table. Use the records attribute to add one or more entries to the table. Because NoSQL is very free-form, a variable number of key-value fields can be appended to the record. The hash_attribute is set to id in the table, so each new record must provide a unique value for the id field.

curl --location --request POST 'http://localhost:9925' \
--header 'Authorization: Basic AuthorizationKey' \
--header 'Content-Type: application/json' \
--data '{
    "operation": "insert",
    "schema": "dev",
    "table": "dog",
    "records": [
        {
            "id": 1,
            "dog_name": "Penny",
            "age": 7,
            "weight": 38
        }
    ]
}'

{"message":"inserted 1 of 1 records","inserted_hashes":[1],"skipped_hashes":[]}

To confirm the record has been added, retrieve the data using an SQL query. To send an SQL query to HarperDB, specify sql for the operation and set sql to the desired SQL statement. The query "SELECT * FROM dev.dog retrieves all records from the table. The output confirms Penny has been added to the table.
Note
NoSQL data is not normalized or columnar, so the key-value pairs do not necessarily appear in any particular order.
curl --location --request POST 'http://localhost:9925' \ --header 'Authorization: Basic AuthorizationKey' \ --header 'Content-Type: application/json' \ --data '{ "operation": "sql", "sql": "SELECT * FROM dev.dog" }'
```
[{"weight":38,"id":1,"dog_name":"Penny","__updatedtime__":1690742615459.453,"__createdtime__":1690742615459.453,"age":7}]
```

Change to the console of the hdb2 node and run the same command. The output should be the same, indicating the record has been replicated to this node.

curl --location --request POST 'http://localhost:9925' \
--header 'Authorization: Basic AuthorizationKey' \
--header 'Content-Type: application/json' \
--data '{
    "operation": "sql",
    "sql": "SELECT * FROM dev.dog"
}'

[{"id":1,"age":7,"__createdtime__":1690742615459.453,"weight":38,"dog_name":"Penny","__updatedtime__":1690742615459.453}]

Confirm replication works in the opposite direction. Using the console for the hdb2 node, add a second entry to the dev.dog table. Increment the id to 2 to ensure it is unique within the table.

curl --location --request POST 'http://localhost:9925' \
--header 'Authorization: Basic AuthorizationKey' \
--header 'Content-Type: application/json' \
--data '{
    "operation": "insert",
    "schema": "dev",
    "table": "dog",
    "records": [
        {
            "id": 2,
            "dog_name": "Rex",
            "age": 2,
            "weight": 68
        }
    ]
}'

{"message":"inserted 1 of 1 records","inserted_hashes":[2],"skipped_hashes":[]}

Return to the first node and retrieve all records from the dev.dog table. The reply should now list two dogs, including the entry added on hdb2. This confirms data is replicating in both directions.

curl --location --request POST 'http://localhost:9925' \
--header 'Authorization: Basic AuthorizationKey' \
--header 'Content-Type: application/json' \
--data '{
    "operation": "sql",
    "sql": "SELECT * FROM dev.dog"
}'

[{"weight":38,"id":1,"dog_name":"Penny","__updatedtime__":1690742615459.453,"__createdtime__":1690742615459.453,"age":7},{"weight":68,"id":2,"dog_name":"Rex","__updatedtime__":1690744053074.6084,"__createdtime__":1690744053074.6084,"age":2}]

More Information

You may wish to consult the following resources for additional information on this topic. While these are provided in the hope that they will be useful, please note that we cannot vouch for the accuracy or timeliness of externally hosted materials.

Compute

Storage

Networking

Databases

Services

Developer Tools

Industries

Pricing

Community

Engage With Us

Search Results

No Results

Filters

How to Create a HarperDB Cluster

What is HarperDB?

Before You Begin

How To Install HarperDB

How To Configure and Initialize the HarperDB Cluster

How to Add and Replicate Data on HarperDB

More Information

Your Feedback Is Important

On this page