An Easier Way To Deploy Cassandra Clusters

November 13, 2014 Ben Laplanche

Pivotal CF customers can now download and install DataStax Enterprise for Pivotal CF using OpsManager, seamlessly deploying a resilient, multi-tenant cluster.

image00

Data Services for Pivotal CF

Pivotal CF provides a stable and scalable platform as a service to build your applications—providing your developers with one-command deployments, support for a wide range of languages, and a high level of scalability and resilience.

However, the majority of applications being built today require some form of data service, whether its a traditional relational database or a highly-available Cassandra cluster to provide a key-value store.

This is why building out a broad ecosystem of data services for application developers to choose from is an important part of the Pivotal CF (PCF) offering to ensure that all of the needs of the application developer are met.

One of the ways we are doing this is through partnering with market leaders for popular data service products, such as with DataStax for their Apache Cassandra offering. We have worked with DataStax to build their Enterprise product into a BOSH release, which can be installed by an Operator onto their PCF platform as an add on service.

“Pivotal CF represents the new developer standard, where developers return their focus to development. For those looking to deploy solutions that include DataStax Enterprise database cluster, it provides a quick and easy platform for developers to get up and running in a few clicks,” said Matt Kennedy, Partner Architect at DataStax.

The DataStax Enterprise service is registered with the Services API so that it appears in the Services Marketplace which is accessible to application developers either through the CLI or the Console.

When developers are now writing their applications, they can easily run `cf marketplace` from their terminal and take their pick from a range of services and associated plans, provided by their PCF Operator for them to use.

No more being locked into the existing relational database, just because its what the rest of the applications use.

Provisioning Distributed Database Systems by Yourself is Hard, Until Now

Building and configuring your own Cassandra cluster from scratch requires some careful thought and orchestration. These considerations can impact your ability to scale and bring your cluster online. For example:

  • IP Management: To build your cluster you need to know the IP addresses of each node. You have to configure these into the `cassandra.yaml` file, along with defining the cluster name, the snitch strategy, which nodes will be seeds and finally the datacenter details.
  • Timing: Accurate timing is critical to maintaining replication and a healthy cluster, so decisions have to be made whether to run your own NTP server as well.
  • Sequencial operations: This is all before ensuring that you bring the seed nodes online first, one at a time, waiting for this to complete before starting the rest and waiting to run `nodetool status` and hoping all is well.

DataStax Enterprise, Built on Apache Cassandra is Now Available on Pivotal CF

DataStax Enterprise for Pivotal CF combines the power of DataStax Enterprise with the flexibility of BOSH on Pivotal CF to deploy a fully-configured Cassandra cluster in your on premise Cloud Foundry environment. This provides a multi-tenant cluster that is perfect for development and testing workloads. Application developers can request an instance which creates their own keyspace in the shared cluster, which can be easily bound to their applications.

Installation of the tile configures a 4-node cluster as default, with BOSH taking care of the IP addresses and configuration files to be used. Seed nodes are automatically configured and started first, using an internal NTP server for all nodes to ensure accurate timekeeping.

Running `nodetool status` shows the healthy 4 node cluster

Running `nodetool status` shows the healthy 4 node cluster

BOSH monitors the health of the Cassandra processes and node VMs, taking corrective action to restart processes or the VM if an error occurs. Combine this with the repair functionality from DataStax OpsCenter and you have an easily deployable and resilient multi-tenant cluster.

To see how easy it is to deploy a multi-tenant cluster, watch this video.

How to Use Cassandra Clusters on Pivotal CF

Using the multi-tenant cluster is as simple as looking at the available plans in the marketplace, creating an instance and binding your application to it.

image02

To create an instance run `cf create-service p-cassandra multi-tenant datastax` to create an instance called `datastax` of the `multi-tenant` plan on the `p-cassandra` service.

image05

`cf services` shows that your instance is created.

Using our sample application on GitHub you can now bind and restage your application to start using your instance.

image01

Using curl you can first create your table called `entries` and then post a key called `foo` with a value of `bar` using POST requests.

image06

To check your entry was created successfully you can use curl to perform a GET request.

image04

For more details check out the readme in the example application.

Get Started and Scale Your Application with NoSQL Database Technology

DataStax Enterprise for Pivotal CF enables operators to easily deploy a Cassandra cluster, utilising the four levels of HA provided by Pivotal CF for your applications and Data Services as well as the repair functionality of DataStax OpsCenter to enable a resilient and healthy cluster.

You can read the documentation for DataStax Enterprise for Pivotal CF to read more about the product features.

Get started today, download and install the tile from Pivotal Network to give your developers more freedom and power to build the next generation applications that power your business.

About the Author

Biography

More Content by Ben Laplanche
Previous
From Sea to Trees, Pivotal Data Science Looks at Climate Change in Acadia National Park: Day 3 Field Report
From Sea to Trees, Pivotal Data Science Looks at Climate Change in Acadia National Park: Day 3 Field Report

In posts one and two, we covered the background of the climate-based research in Acadia National Park, the ...

Next
Sprint vs. Iteration
Sprint vs. Iteration

At Pivotal we have a weekly Iteration Planning Meeting, or IPM for short. Often when I explain what happens...