Create a Cluster

To create a cluster click on the + New Cluster button in the Clusters page. You will see the following form:

Create a new cluster

  • Cluster Name: Give an alphanumeric name to your cluster.

  • Cloud Provider: Select your cloud provider. As of 1.0-Beta version, only AWS is supported. We will be supporting Microsoft Azure and Google Cloud Platform in future releases.

  • Region: Select your region. It is important to choose the same region as your client application to achieve low network latency.

  • Cluster Type: Choose a type which defines the size of your cluster. Except for Free, you can specify a size using the "Cluster Memory" slider after you choose a type. You can change this size while your cluster is running without any downtime. Available types are as follows:

    • Free: Free clusters have a fixed memory size of 200 MB. They also have CPU limitations, so performance will be lower than you’ll have at a paid tier. Free clusters are automatically stopped after being idle for 16 hours. This cluster type is only recommended for development, testing, or small applications.

    • Small: Small clusters have a memory range of 2 GB to 15 GB. They can be scaled up or down, within that range, by intervals of 1 GB.

    • Medium: Medium clusters have a memory range of 15 GB to 75 GB. They can be scaled up or down, within that range, by intervals of 5 GB.

    • Large: Large clusters have a memory range of 75 GB to 375 GB. They can be scaled up or down, within that range, by intervals of 25 GB.

The memory sizes mentioned in Cluster Types denote the total reserved memory for your map data including any primary and backup replicas. You need to take this into account when doing size calculations. For example, if you have a Small cluster with 2 GB memory capacity and a map backup count set to 1, this means that you can insert up to 1 GB of data (1 GB primary + 1 GB backup).
  • Enable Auto Scaling: When enabled, hazelcast cluster will grow and shrink automatically depending on memory utilization. When used memory percentage is over 80%, cluster is scaled up, when used memory percentage is lower than 40%, cluster is scaled down. You can enable/disable this option while the cluster is running too.

  • Enable Persistence: When you enable persistence, you do not lose data after you shutdown the cluster. Using the Hot Restart feature of Hazelcast Enterprise, the cluster data is stored on the disk and read back into the cluster upon restart. Enabling persistence can degrade the throughput up to 40%, so take that into consideration along with the specifics of your use case to decide whether this is a good trade off.

  • Enable Encryption: When enabled, Hazelcast Cloud will use TLSv1.2 with mutual authentication to secure and encrypt the communication between Hazelcast members and clients. Enabling encryption has a minimal impact on the performance and throughput of your cluster.

  • Enable IP Whitelist: By default any client with correct credentials can connect to your cluster. You may want to restrict this by defining IP white list. If you enable IP white list then only the clients with the listed IP addresses will be able to connect your cluster.

  • Custom Map Configuration: You can configure your Hazelcast map using the parameters shown in the following screenshot. If you leave the map name as default, the configuration will apply to all maps. Otherwise, only the map with the specified name will use the configuration. You can also apply wildcards to the map name. For more information on configuring with wildcards, see Using Wildcards.

Configuration options for maps

For information about custom map configuration options, see Map Configurations

Currently, Hazelcast Cloud only supports the Map Data Structure. Other data structures can be used, but their total size should not exceed 50 MB, as these will be placed on heap and we do not currently provide heap configuration.
For security reasons, dynamic classloading is disabled in Hazelcast Cloud. This means that you cannot use Hazelcast’s Executor Service or Entry Processor features. Also, if you are planning to query the custom objects inside your map, you should use Portable Serialization for your custom objects. For more information about Portable, see Implementing Portable Serialization. We are planning to add a secure way to load your own classes to the Hazelcast Cloud clusters in a future release.