We consider defenses against confidentiality and integrity attacks on data following break-ins, or so-called intrusion resistant storage technologies. We investigate the problem of protecting secret data, assuming an attacker is inside a target network or has compromised a system. We give a definition of the problem area, and propose a solution, VAST, that uses large, structured files to improve the secure storage of valuable or secret data. Each secret has its multiple shares randomly distributed in an extremely large file. Random decoy shares and the lack of usable identification information prevent selective copying or analysis of the file. No single part of the file yields useful information in isolation from the rest. The file’s size and structure therefore present an enormous additional hurdle to attackers attempting to transfer, steal or analyze the data. The system also has the remarkable property of healing itself after malicious corruption, thereby preserving both the confidentiality and integrity of the data.