Understanding persistence is important when you evaluate different data store systems. Making the wrong choice or taking the wrong approach could mean substantial downtime or data loss. 

Let’s look at data persistence in the context of Apache Cassandra®.

If you’re interested in learning more about persistence in Cassandra and other NoSQL databases, check out our complete guide to NoSQL.

What is data persistence?

Persistence is "the continuance of an effect after its cause is removed." Persistent data is stored on a long-lasting storage medium, and data remains intact after modifications or changes to the storage medium. 

This is the important distinction between persistent data and non-persistent data: the data survives after the process ends. In other words, for data to be stored persistently, it must write to non-volatile storage.

Data persistence ensures that valuable business information remains accessible in a database and consistent across sessions, devices, and applications. 

Data stored could be

• transactional data (dynamic data) on ecommerce platforms 

• customer records (sensitive data)

• user-generated content 

The point of saving data in a non-volatile storage system is to make sure it can be reliably retrieved later. Persistent storage protects data integrity—it’s efficient data management.

Which data stores provide persistence?

If you need persistence when storing data, there are four main design approaches:

• Pure in-memory storage—no persistence at all (such as memcached or Scalaris)

• In-memory with periodic snapshots (such as Oracle Coherence or Redis)

• Disk-based with update-in-place writes ( such as MySQL ISAM or MongoDB)

• Commitlog-based (such as all traditional OLTP databases like Oracle, SQL Server, etc.)

Robust data management capabilities, such as backup and recovery, indexing, and data replication, are critical for ensuring data consistency,  security, and accessibility across various applications.

Adding persistence to systems

In-memory (not persistent data)

In-memory approaches can achieve blazing speed, but at the cost of being limited to a relatively small data set. Most workloads have relatively small "hot" (active) subset of their total data; systems that require the whole dataset to fit in memory rather than just the active part are fine for caches but a bad fit for most other applications. Because the data is in memory only, it will not survive process termination. Therefore these types of data stores are not considered persistent.

Snapshots

The easiest way to add persistence to an in-memory system is with periodic snapshots to disk at a configurable interval. But you can lose up to that interval's worth of updates.

Different methods to persist data include local and remote storage solutions, each with varying levels of control and privacy over saved information. 

Update-in-place and commitlog

Update-in-place and commitlog-based systems store to non-volatile memory immediately, but only commitlog-based persistence provides Durability -- the D in ACID -- with every write persisted before success is returned to the client.

Can you achieve data persistence with a database?

Yes! Databases are a common—and effective—way to achieve data persistence, whether using a relational database or a NoSQL database like Cassandra.

Cassandra data persistence

Cassandra implements a commit-log-based persistence design but, at the same time, provides for tunable levels of durability. This allows you to decide what the right trade-off is between safety and performance. For each write operation, choose to wait for that update to be: 

• buffered to memory

• written to disk on a single machine

• written to disk on multiple machines

• written to disk on multiple machines in different data centers 

Or, you can accept writes as quickly as possible, acknowledging their receipt immediately before they have even been fully deserialized from the network.

Data persistence with Astra DB

Astra DB is a cloud-native, fully managed version of Cassandra that includes additional capabilities and optimizations for cloud environments, infrastructure management, scaling, and maintenance tasks. It ensures data persistence in a number of ways.

Key benefits 

Consistent data storage: Astra DB ensures data remains consistent and durable across distributed environments, maintaining integrity even during network partitions or node failures.

Automated backups: Regular automated backups with configurable retention policies safeguard data against accidental loss or corruption.

Point-in-time recovery: Allows data restoration to a specific moment, enhancing data recovery options and minimizing potential data loss.

Replication: Automatic data replication across multiple nodes and data centers ensures data persistence and availability, even in case of hardware failures.

Commit log: Writes are first recorded in a commit log, ensuring the durability of data even in the event of sudden system failures.

Tunable consistency: Configurable consistency levels for read and write operations allow balancing between data persistence guarantees and performance needs.

Why data persistence matters

You decide what the right performance/durability trade-off is for your data. Making an informed decision on how you store data is critical to addressing this tradeoff—on your terms. 

Data persistence ensures that data remains consistent and intact, safeguarding information from unauthorized modifications and maintaining data integrity across sessions and applications. 

Because Cassandra provides such tunability, it is a logical choice for systems with a need for a durable, performant data store.