Nowadays Cassandra is getting a lot of attention, and we’re seeing more and more examples of companies moving to Cassandra. Why is this happening? Why are companies with solid IT structures and internal knowledge shifting, not only to a different paradigm (Read: NoSQL vs SQL), but also to completely different software? Companies don’t simply move to Cassandra because they feel like it. A drive or need must exist. In this post, I’m going to review a few use cases and highlight some of the interesting parts to explain why these particular companies adopted Cassandra. I will also try to address concerns about Cassandra in enterprise environments that have critical SLAs and other requirements. And at the end of this post, I will go over our experience with Cassandra.
Cassandra Use Cases
Cutting costs. How? Instagram was using an in-memory database before moving to Cassandra. Memory is expensive compared to disk. So if you do not need the advanced performance of an in-memory datastore, Cassandra can deliver the performance you need and help you save money on storage costs. Plus, as mentioned in the use case, Cassandra allows Instagram to continually add data to the cluster. They also loved Cassandra’s reliability and availability features.
Cassandra proved to be the best technology, among the ones they tested, for their scaling needs. With Cassandra, Ebay can look up historical behavioral data quickly and update their recommendation models with low latency. Ebay has deployed Cassandra across multiple data centers.
Spotify moved to Cassandra because it’s a highly reliable and easily scalable datastore. Their old datastore was not able to keep up with the volume of writes and reads they had. Cassandra’s scalability with its multi-datacenter replication, plus its reliability, proved to be a hit for them.
They were looking for 3 things: scale, availability, and active-active. Only Cassandra provided all of them. There transition to Cassandra went smoothly and enjoy the ease of development Cassandra offers.
Cassandra brings something new to the game
NoSQL existed before Cassandra. There were also other mature technologies when Cassandra was released. So why didn’t companies move to those technologies?
Like the subtitle says, Cassandra brings something new to the game. In my experience, and as discussed in some of the use cases above, one of the strongest points is Cassandra’s ease of use. Once you know how to configure Cassandra, it’s almost “fire-and-forget”! It just works. In an era like ours, where you see new technologies appear every day, on different stacks, with different dependencies, Cassandra easy installation and basic configuration is refreshingly simple, which leads us to…
Scalability!! Yes it scales linearly. This level of scalability, combined with its ease of deployment, takes your infrastructure to another level.
Last but not least, Cassandra is highly flexible. You can tweak your consistency settings per transaction. You need more speed? Pick less consistency. You want data integrity? Push those consistency settings up. It is up to you, your project, and your requirements. And you can easily change it.
Also don’t forget its other benefits: open source, free, geo-replication, low latency etc…
Pythian’s Experience with Cassandra
Cassandra is not without its challenges. Like I said earlier, it is new technology that makes you think differently about databases. And because it’s easy to deploy and work with, it can lead to mistakes that could seriously impact scalability, and applications/services performance when they start to scale.
And that is where we come in. We ensure that companies just starting out with Cassandra have well built and well designed deployments, so they don’t run into these problems. Starting with a solid architecture plan for a Cassandra deployment and the correct data model can make a whole lot of difference!
We’ve seen some deployments that started out well, but without proper maintenance, fell into some of the pitfalls or edge-cases mentioned above. We help out by fixing the problem and/or providing recommended changes to the original deployment, so it will keep performing well without without issues! And because Cassandra delivers high resilience, many of these problems can be solved without having to deal with downtime.
Thinking about moving to Cassandra? Not sure if open source or enterprise is right for you? Need project support? Schedule a free assessment so we can help you with next steps!
I have two doubts.
. How can I get started with Cassandra if I am an Oracle DBA? I mean, what kind of training do I have to take?
.. Cassandara or Hadoop?
To start with Cassandra Datastax academy provides free online training and it is one of the best resources available: https://academy.datastax.com/
Cassandra and Hadoop aren’t mutually exclusive. It quite usual to see Hadoop environments with Cassandra.
Thanks Carlos. I’m going to take your advice.