4C - Collaboration to Clarify the Costs of Curation

PrestoPRIME Cost Model for Digital Storage (PP-CMDS)

The PrestoPRIME model and tools were developed within an EC project and provide information for forecasting the costs of long-term preservation of mass digitised AV materials and comparing the costs of different preservation solutions.

The tools can be used for modelling and simulating the costs and risks of using IT storage for the long-term archiving of file-based audio visual assets. The objective is not to provide exhaustively detailed models or highly accurate costs but to provide simple results that are meaningful and useful in specific contexts, for example when selecting storage strategies or supporting day-to-day decision making. It accounts for cost as monetary expenses and risks of loss of assets.

The total annual cost of managed storage per copy is approximately 10 times today’s cost of commodity storage media. The lifetime cost of managed storage per copy is approximately 4 times the current total annual cost of managed storage.

The cost levels incorporated in the model are based on experience from storage providers and a few large scale institutions. They conclude that cost-effective long-term storage at any scale requires automation through mass storage devices (disk servers and tape robots). The short obsolescence times for the components of these systems, including media, require the use of continual migration and active management.

The long-term storage planning tool allows the user to define storage media types (called storage systems), combine them in storage configurations and use them with collections. Media types can be added with storage migration period, running storage cost per volume unit per year with increasing or decreasing cost. The same can be done for running access cost. Corruption rates, latent and on access can be entered. Storage configurations are combinations of storage media types with given access share and frequency—and scrubbing. Collections can be added with average file size, length of cost/loss projection, amount of file and rate of change. The final plan consisting of a collection and a storage configuration can then be evaluated, with the result showing as three graphs over time, showing risk and loss, costs and corruption and loss.

The Storage simulation tool named iModel is a desktop GUI application, running on Java. It simulates the work environment of an archive manager by requesting that they take action as time goes by, requiring that they allocate resources where most needed. Actions can be file ingest, integrity checks, file repair using an alternative file copy, file access and so on. It has the same elements as the long-term planning tool, and can also simulate catastrophic events, such a losing all data on one set of media.

Tools for higher level preservation modelling, for example file format migration, digitisation and cataloguing that apparently were under development have not been released.