Apache Cassandra is the easiest truly big-data database that can scale and replicate data globally in a master-less configuration. What used to be in the hands of only the biggest in Silicon Valley is now available as a mature database to the masses. Originally created at Facebook after they studied Amazon's DynamoDB and Google's BigTable whitepapers, the Cassandra we know today is very different and has far surpassed its ancestors in feature set and has now become a popular wire-protocol for other databases such as ScyllaDB, YugaByte, and Azure's CosmosDB.
Apache Cassandra is written in Java language. I think why it is chosen to be written in Java may be because the security is a prime concern it is developed in Java rather than in C++. Another key reason could be Performance. It might be slower at the startup, but once the code is ready and in running state it is way faster as compared to C++. Java code is continuously optimized by the JVM and in that consideration, it appears faster to C++. It may have other reasons as well such as advanced memory optimization or efficient garbage collection.
There are different flavors of Apache Cassandra available in the market,
- ScyllaDB is an open-source distributed NoSQL datastore which was intended and designed with Apache Cassandra while achieving expressively higher throughput and lower latencies. It’s written in C++.
- YugaByte DB is a transactional and high-performance distributed database for building largescale scattered cloud services. It also supports APIs which are Cassandra compatible and Redis compatible, with PostgreSQL in the Beta stage.
YugaByte DB core is written in C++, but the repository contains Java-based code that is needed to run sample applications.
- DataStax Enterprise offers Apache Cassandra flavor in a database platform which is built knowingly for the purpose of providing performance and availability demands of IOT, Web and Mobile applications. It gives organizations a safe always-on database that effects operationally simple when scaled in a single or across multiple data centers and in the clouds. Cassandra and DataStax Enterprise have helped the customers supporting multi-datacenter and hybrid cloud deployments since the beginning. It is written in Java.
Let’s quickly jump to the merits of using Apache Cassandra!
Cassandra is a fantastic platform while handling amounts of unstructured data at scale. If you’re struggling with making your relational database faster and more reliable—mostly when you’re at scale—Cassandra may be considered as a great option for you. It unites the Amazon’s Dynamo storage system along with Google’s Bigtable model, offering the near-constant availability required to support real-time querying for web and mobile apps.
- Cassandra can handle even the most massive datasets.
- It can work as amazing, record-setting reliability at scale.
- Eventual consistency yields high availability.
- It offers Wide-column flexibility.
- It also offers minimal administrative tasks at scale.
- It offers easy setup and maintenance (does not matter how big the dataset that you are setting)
- Flexible parsing and wide column requirements.
- Not with multiple secondary indexes.
- It allows applications to write into any node anywhere and anytime.
- Automatic workload management and data balancing across the nodes
- Linearly scalable by just adding more nodes to the cluster.
On the other hand, it may not be a proven benefit if,
- Your app requires transactional operations,
- If requires dealing with financial data,
- Need dynamic queries against column data,
- Low latency requirement,
- Read exceeds write by a large margin
- On-the-fly aggregations & joins and so on...
It is hard to find which large-scale organization does not use Cassandra nowadays. When dealing with distributed databases, it is always the key requirement to identify how the data and the workload will be distributed. Correspondingly, the data model must be correctly designed. For example,
- Not letting the partition key too large,
- A specific size of the tables,
- Keeping an ideal same partition size, etc.
The most important point to highlight is even though distributed databases falls under the category of the database, however, treating this application to behave like a traditional relational database may incur excessive performance degradation and it may break the application as well. So, we must have to be careful while designing the application.
Trackbacks/Pingbacks