* Fixed content that needs to be highly available
Backing up and recovering data has always been the one function most closely identified with IT staff. Over the years, we have learned to capture, as nearly as possible, accurate representations of information at every point where it is modified.
The result is that we provide our users with recovery points that can be extremely close to what they think they need. In many instances – even those cases in which users really have no functional memory of when the file they want was last in the state in which they now need it to be – we have spent a lot of money to ensure that we can usually provide users with an effective recovery.
Throughout the roughly 50 years that data has been written to disk, IT personnel have mostly been interested in when data changes largely because of the need to provide recoveries.
However, some data – even though it gets used on a daily basis – does not change.
“Fixed content” is information that never changes once it has been created. Examples of fixed content include reference data such as MRIs, X-rays, CAT scans and other digitized medical records; bank records of scanned checks and signatures; and the digitized histories that brokerages must keep on many of their clients’ transactions.
Sometimes fixed content is “fixed” because of regulatory requirements (HIPAA and Sarbanes-Oxley, for example). Sometimes it is fixed because an organization needs to refer to a common data set over an extended period of time.
Whether fixed by regulation or simply by need, in each case this reference data is expected to be maintained in an unchanged, immutable state. In some respects, fixed content sounds like an archive.
Can fixed content be treated the same way as a standard archive? No way.
Whereas archived data is rarely accessed, fixed content must be thought of as reference data. And like the reference data in a dictionary, we should expect it to be accessed repeatedly.
Because reference data must be highly available, logic dictates that it be kept online rather than archived to tape. And because the content must not change, expect WORM (Write Once, Read Many) capability to be another requirement.
When the fixed content represents a large amount of information – a likely scenario – it becomes increasingly important that the data repository have some highly optimized method for identifying and accessing the data. A few years ago when EMC introduced the Centera, (and by doing so, giving the concept of fixed content its first real exposure) it described such storage as being “content-addressable,” which meant a content-based system of addressing the data.
More recently, Archivas has introduced a clustered solution with an object-based system for managing such data.
The Archivas approach aggregates data, metadata and management policies, enabling applications to retrieve objects and not files. The company provides a completely open platform that has been designed to be both self-managing and serviceable as a single repository for multiple applications.
