What is Data at Rest?

When data is not being transmitted between locations, it is considered to be at rest. In its rawest form, data is stored in digital form (represented as binary 0 and 1 values) on storage media, which can reside on-premise, on portable media, or in the cloud. Examples of portable storage include a hard disk drive in a powered enclosure, tape cartridges, a flash memory stick, CF or SSD card. Cloud storage subscriptions are priced based on capacity and speed. High-speed storage leverages Solid-State Drives (SSD); spinning hard disks is slower but costs less per Megabyte of storage.

Data Volatility

For data to be changed by software applications or utility programs, it must be read into short-term, volatile memory such as RAM or a CPU cache. The term “data at rest” is usually applied to digital information that resides on non-volatile storage media. Early forms of non-volatile data storage included paper tape, punched cards, and floppy disks. Magnetic tape is still used for off-site archiving at disaster recovery centers.

Sequential Versus Direct Access Media

Data stored on magnetic tape can only be read sequentially. Data stored on a disk can be read at random data blocks, but it can take several milliseconds for the reading arm to seek the right track, and you need to wait for your data block to spin to be under the reading head, which is known a rotational delay. This slight delay is the primary reason spinning disks are slower than SSDs, which have no moving parts.

Price Performance Considerations

Stored data is subject to a price-to-performance tradeoff. The highest performance storage is required to meet the needs of real-time applications where milliseconds of delay can result in lost business revenue. Slower storage, such as that used by a data lake, is usually stored on less costly mass storage. As the value and frequency of access to data declines, it can be moved to archive media, which can be online, nearline, or offline (hot, warm, cold).

Data at Rest and Matching Applications to Storage Types

Data at rest is organized in a way that makes using it efficient. Large volumes of data can be stored in the data lake, data warehouse, or used by online transactional processing (OLTP) systems. Hadoop clusters and AWS S3 buckets are common repositories for data lakes. Data warehouses are focused on returning query results and are commonly organized for access by a relational database system (RDBMS). To scale to larger data volumes, data can be shared with multiple servers in a server cluster that enables queries to be broken down into smaller units that then can be run in parallel across multiple servers. Clusters can share stored data by federating a data set across multiple servers or by using shared-everything architectures.

Data at Rest and Security

In an RDBMS, data at rest encryption allows all columns in all tables in the database to be encrypted. Data in the encrypted database is stored on disk or other media in encrypted form and can only be accessed if the encryption key is known. Encrypted columns are often stored in database files using 256-bit Advanced Encryption Standard (AES) encryption. The encryption is transparent to the applications accessing the data.

How Actian Data Management Use Data

Actian data management solutions use a federated data architecture for its clustered systems and increasingly use cloud storage that decouples storage from compute resources to ease management and allow customers to subscribe to only the amount of storage and compute that their applications need.

Data at rest and data in motion use encryption to ensure data security. The Actian Zen DBMS stores its data in a way that allows both relational and key-value access methods. Actian Ingres supports hybrid OLTP and data warehouse applications, so the data at rest formats stored are both traditional row-based RDBMS format to support OLTP applications and in a columnar format to optimize data warehouse use cases that can use parallel query execution.

The Actian Data Platform is a hybrid cloud data management solution that is cluster-aware and supports the spark API for Hadoop file formats and cloud storage, including AWS S3, Azure Blob storage and Google Cloud storage. You can learn more about the Actian Data Platform here.