Launching a new Cluster with the Kognitio Launcher on AWS

Launching a new cluster creates one or more EC2 instances and installs a Kognitio server on it / them.

Prerequisites

Create a Kognito Cluster on AWS

1. Enter your Cluster Details

Fill out or select the following:

  • Name - This is the name of your new cluster and will be used in the name of any resources created as part of the cluster creation.

  • Admin Login - Either select previously created credentials that you want to use or keep the default of “Create New Credentials” to create a new set of credentials. If in any doubt, use the default. Credentials control how you connect to your cluster and are described in Manage Credentials

  • Launch Template - The launch options provide fine control over how the EC2 instances are created and is an advanced setting. Unless you have specific requirements, choose “Default Launch Options”. Launch Options are described in Manage Launch Templates

The “Show Launch Settings” button allows you to view the launch settings for the cluster you are about to create. See Launch Settings for more details.

Click “Next (Instance Settings)”.

2. Instance Settings

There are 2 ways to specify the type and number of EC2 instances to use for your cluster. If you do not know exactly what resource you need then we recommend you make use of the Calculate from data size/usage tab. If you know what resource you want then use the Specify Instance Details tab.

Calculate from data size/usage

The Launcher will automatically calculate the size of the cluster to create based on the following:

  • Usage level - choose your expected usage from the guidelines below:

    Usage

    System type

    Light

    Optimises for lowest cost using memory optimised instances. Suitable for a single power user or a small number of users running moderately complex SQL

    Standard

    Several concurrent users running complex SQL over large proportions of the data or advanced analytics

    Heavy

    Higher concurrency complex SQL and/or more demanding advanced analytics (large R models etc). Tens of concurrent BI tool users

    Very Heavy

    100s of concurrent users. Demanding performance requirements. Large scale reporting and advanced analytics environment

  • Data Size - Specifiy the size of your data you want to hold in memory to the nearest Terabyte (Ti) or Gigabyte (Gi).

  • Persistent Data Size - Specify the size of the data that you want to store in Kognitio persistent storage (internal tables).

    If you are using a data store such as S3, HDFS etc to store all your data and simply imaging that data in Kognitio memory then your Persistent Data Size can be set to 0.

    If you want to store data in Kognitio internal tables then enter the size here. It is better to overestimate as disk storage is relatively cheap in AWS EC2 instances.

As you alter the 3 parameters outlined above the instance type and count are changed to reflect your selection. The cluster size and Guide Price ($/hour) are shown below the selected configuration.

AWS Instance calculation based on standard usage and 500 GiB data in RAM

Specify Instance Details

This tab allows you to specify the type and number of EC2 instances to build your cluster from. Select the “Calculate from data size/usage” tab above if you are unsure what to choose.

  • Instance Type - this is the EC2 instance type to use for the cluster. Instances with Memory sizes below 16GB will have restricted external connector and external script performance and we recommend 64GB+ of memory for best performance.

  • Instance Count - this is the number of EC2 instances you want in your cluster. It is better to have a small number of large instances than a lot of small ones.

  • Disk Size - this is the amount of disk for each node. Even if you don’t want to store data in Kognitio internal tables you should allocate at least 16GB for the system tables and logs.

  • Request Spot - this launches the EC2 instances on spot. This is cheaper than on demand instances but the instances can disappear at any time taking any data in internal or RAM only tables with them. Because of the distributed nature of a Kognitio cluster, losing one instance will break the whole cluster.

Once you have entered your cluster specification, click on Next (Review and Submit)

3. Review and Submit

Review the specifications of the AWS instances you are about to launch and enter a SYS Password which you can use to connect to the main Kognitio administration account (“sys”) using Most tools with JDBC or ODBC connectivity.

Click on Launch Cluster to submit the request to AWS.

4. Next Steps

The time taken to launch Kognitio will depend on the size of your cluster but should only be a few minutes. As the cluster is created you will see progress messages and once the cluster is ready you will be presented with the “Management” screen that gives an overview of the resources available on your cluster.

Kognitio cluster Management Screen

Your Kognitio cluster is now ready to use: