DB/Nosql
MongoDB vs Casandra
데먕
2019. 9. 30. 09:31
1. Overview
When it comes to NoSQL databases, MongoDB and Cassandra may seem similar but are quite different. Both have similar use cases but are not intended for transactional data (i.e., accounting systems)
2. Description
2.1 Comparison of Features
| Feature | MongoDB | Cassandra |
| Development language | C++ | Java |
| Storage Type | BSON files(Document-based) | Column(Wide-columns) |
| Expressive object model | Yes | No |
| Protocol | TCP/IP | TCP/IP |
| Secondary Indexes | Yes | No |
| Transactions | No or Yes(Support for multi-document ACID transactions was added to MongoDB with the General Availability of the 4.0 release) | Local |
| Performance | Provides faster query execution performance due to simple schema | Provide high scalable and linear performance |
| Fault tolerance | Master-slave model + failover, with no single point of failure | High availability without a single point of failure |
| No downtime on node failure | No | Yes |
| Scalability | Writing scalability is limited in MongoDB | High scalability |
| Speed | Not faster compared to Cassandra in terms of read operations | Faster for queries of smaller scripts |
| Query Speed | Reduces network speed while multi-query execution | Multi-Query execution capacity |
| Read/Write capabilities | Meets low latency requirements for reading requests, at low throughput | Meets low latency requirement for write requests, at low throughput. Not engineered for reading performance |
| High write throughput | No | Yes |
| Sharding | Sharding is built-in. But vs Cassandra, MongoDB shards are replications of multiple servers | Auto-sharding, with a different effect on its replication. In Cassandra, each shard is a server, with its data replicated across other servers |
| Querying | "Query by example" model. Mongo can execute complex, dynamic queries. but it's not necessarily the best fit for reporting-style workloads or complex transactions | Predictable query performance, but being a key/value store it can be a bit more tricky than MongoDB. it allows querying by key or key range |
| Concurrency | Instant update | MVCC |
| Locks | Yes | Yes |
| Triggers | No | Yes |
| Replication | Master-Slave. MongoDB's replication facility, called replica set, provides automatic failover and data redundancy. Replication needs to be set up, but Mongo has clear guidelines to help | Multi-Master. Replication with Cassandra is very simple. It mostly takes care of it all for you. Just indicate how many nodes you want your data copied to and it automatically handles the rest. |
| CAP theorem | Consistency, Partition tolerance | Partition tolerance, High Availability |
| Operating System | Linux, Mac OS, Windows | Linux, Mac OS, Windows |
| Data storage | Disc | Disc |
| Characteristics | Retains some SQL properties such as query and index | A cross between BigTable and Dynamo. High availability |
| Hadoop support | MongoDB Connector for Hadoop | Native Hadoop support, including Hive and Pig |
| Data Integration | Fast and Flexible with different applications and tools such as Big Data | Fast and Flexible with different tools |
| Pricing | Open source and depends on the implementation and has different licensed versions too | Open source and depends on the implementation |
| Syntax | Requires you learn a new query language | CQL is easier to learn alike SQL than MongoDB |
| Areas of use | CMS system, comment storage | Banking, finance, logging |
3. References
https://www.upwork.com/hiring/development/mongodb-vs-cassandra/