Content-addressable storage

Content-addressable storage, also referred to as associative storage or abbreviated CAS, is a mechanism for storing information that can be retrieved based on its content, not its storage location. It is typically used for high-speed storage and retrieval of fixed content, such as documents stored for compliance with government regulations. Roughly speaking, content-addressable storage is the permanent-storage analogue to content-addressable memory.

CAS and FCS

Content Addressable Storage (CAS) and Fixed Content Storage (FCS) are different acronyms for the same type of technology. The CAS / FCS technology is intended to store data that does not change (fixed) in time. The difference is that typically CAS exposes a digest generated by a cryptographic hash function (such as SHA-1 or MD5) from the document it refers to. If the hash function is weak, this method could be subject to collisions in an adversarial environment (different documents returning the same hash). The main advantages of CAS / FCS technology is that the location of the actual data and the number of copies is unknown to the user.

Content-addressed vs. location-addressed

When being contrasted with content-addressed storage, a typical local or networked storage device is referred to as location-addressed. In a location-addressed storage device, each element of data is stored onto the physical medium, and its location recorded for later use. The storage device often keeps a list, or directory, of these locations. When a future request is made for a particular item, the request includes only the location (for example, path and file names) of the data. The storage device can then use this information to locate the data on the physical medium, and retrieve it. When new information is written into a location-addressed device, it is simply stored in some available free space, without regard to its content. The information at a given location can usually be altered or completely overwritten without any special action on the part of the storage device.

Within the scope of this discussion, a good way to think of the above is as container-addressed storage.

The Content Addressable File Store (CAFS) was a hardware device developed and sold by International Computers Limited (ICL) in the 1970s and 1980s that provided location-addressed disk storage with built-in search capability. The search logic was incorporated into the disk controller. A query expressed in a high-level query language could be compiled into a search specification that was then sent to the disk controller for execution. Files could also be accessed via the conventional location-addressing mechanism, permitting CAFS to support an IDMS CODASYL database and also support content addressing of the same records.

In contrast, when information is stored into a CAS system, the system will record a content address, which is an identifier uniquely and permanently linked to the information content itself. A request to retrieve information from a CAS system must provide the content identifier, from which the system can determine the physical location of the data and retrieve it. Because the identifiers are based on content, any change to a data element will necessarily change its content address. In nearly all cases, a CAS device will not permit editing information once it has been stored. Whether it can be deleted is often controlled by a policy.

While the idea of content-addressed storage is not new, production-quality systems were not readily available until roughly 2003. In mid-2004, the industry group SNIA began working with a number of CAS providers to create standard behavior and interoperability guidelines for CAS systems.

Pros and cons

CAS storage works most efficiently on data that does not change often. It is of particular interest to large organizations that must comply with document-retention laws, such as Sarbanes-Oxley. In these corporations a large volume of documents will be stored for as much as a decade, with no changes and infrequent access. CAS is designed to make the searching for a given document content very quick, and provides an assurance that the retrieved document is identical to the one originally stored. (If the documents were different, their content addresses would differ.) In addition, since data is stored into a CAS system by what it contains, there is never a situation where more than one copy of an identical document exists in storage. By definition, two identical documents have the same content address, and so point to the same storage location.

For data that changes frequently, CAS is not as efficient as location-based addressing. In these cases, the CAS device would need to continually recompute the address of data as it was changed, and the client systems would be forced to continually update information regarding where a given document exists. For random access systems, a CAS would also need to handle the possibility of two initially identical documents diverging, requiring a copy of one document to be created on demand.

Further information: Copy on write

Typical implementation

Paul Carpentier and Jan van Riel coined the term CAS while working at a company called FilePool in the late 1990s. FilePool was acquired in 2001 and became the underpinnings of the first commercially available CAS system, which was introduced as EMC's Centera platform. Paul and Jan are now working together again at Caringo which has introduced advancements in CAS technology with the CAStor content storage software. The Centera CAS system consists of a series of networked nodes (1-U servers running Linux), divided between storage nodes and access nodes. The access nodes maintain a synchronized directory of content addresses, and the corresponding storage node where each address can be found. When a new data element, or blob (Binary large object), is added, the device calculates a hash of the content and returns this hash as the blob's content address. As mentioned above, the hash is searched for to verify that identical content is not already present. If the content already exists, the device does not need to perform any additional steps; the content address already points to the proper content. Otherwise, the data is passed off to a storage node and written to the physical media.

When a content address is provided to the device, it first queries the directory for the physical location of the specified content address. The information is then retrieved from a storage node, and the actual hash of the data recomputed and verified. Once this is complete, the device can supply the requested data to the client. Within the Centera system, each content address actually represents a number of distinct data blobs, as well as optional metadata. Whenever a client adds an additional blob to an existing content block, the system recomputes the content address.

To provide additional data security, the Centera access nodes, when no read or write operation is in progress, constantly communicate with the storage nodes, checking the presence of at least two copies of each blob as well as their integrity. Additionally, they can be configured to exchange data with a different, e.g. off-site, Centera system, thereby strengthening the precautions against accidental data loss.

IBM has another flavor of CAS which can be software based, Tivoli Storage manager 5.3, or hardware based, the IBM DR550. The architecture is different in that it is based on a hierarchical storage management (HSM) design which provides some additional flexibility such as being able to support not only WORM disk but WORM tape and the migration of data from WORM disk to WORM tape and vice versa. This provides for additional flexibility in disaster recovery situations as well as the ability to reduce storage costs by moving data off disk to tape.

Another typical implementation is iCAS from iTernity. The concept of iCAS is based on containers. Each container is addressed by its hash value. A container holds different numbers of fixed content documents. The container is not changeable and the hash value is fixed after the write process.

Open-source implementations

One of the very first content-addressed storage servers, Venti, was originally developed for Plan 9 from Bell Labs and is now also available for Unix-like systems as part of Plan 9 from User Space.

A first step towards an open source CAS+ implementation is Twisted Storage.

Tahoe Least-Authority File Store is an open source implementation of CAS.

While it is generally used as a source code control system, Linus Torvalds' Git program is a userspace CAS filesystem.

Project Honeycomb is an open source API for CAS systems.

The XAM interface being developed under the auspices of the Storage Networking Industry Association is an attempt to create a standard interface for archiving on CAS (and CAS like) products and projects.

Bitcache is an open source distributed implementation of CAS written in Ruby. Bitcache server has an implementation for Drupal as well.

Camlistore is a recent project to bring the advantages of content-addressable storage "to the masses". It is intended to be used for a wide variety of use cases, including distributed backup; a snapshotted-by-default, version-controlled filesystem; and decentralised, permission-controlled filesharing.

Irmin is an ocaml "library for persistent stores with built-in snapshot, branching and reverting mechanisms"; the same design principles as Git.

Cassette is an open source CAS implementation for C#/.NET.

Arvados Keep is an open source content-addressable distributed storage system. It is designed for large-scale, computationally intensive data science work such as storing and processing genomic data.

InterPlanetary File System (IPFS), is a content-addressable, peer-to-peer hypermedia distribution protocol.

See also

• Content Addressable File Store
• Content-centric networking / Named data networking
• Data Defined Storage

External links

• Fast, Inexpensive Content-Addressed Storage in Foundation
• Venti: a new approach to archival storage