Although increasingly rare, disk corruptions can occur due to a variety of unique causes. Whether caused by hard disk or transient memory errors, corruptions can occur within the file system’s metadata, which links physical disk blocks to virtual data. To restore access to the corrupted data, Windows must isolate and correct the issue using the chkdsk tool. Windows 8 vastly improves the resolution time of system errors by implementing a new NTFS health model and Check Disk resolution strategies.
Key changes of the design
In previous versions of the Windows operating system, NTFS implemented a simple health model which indicated two states for a volume; healthy or not. In this health model, the corrupted volume had to be taken offline for however long was necessary to resolve the issue. The downtime of the volume was directly proportional to its number of files, which for some systems was quite large. Windows Vista and Windows 7 made significant improvements to this process that ultimately decreased the resolution time. However, with hard drive capacities rapidly expanding, using the current resolution method has spanned several hours in some cases.
A redesigned NTFS health model and chkdsk was necessary to correlate with the rapidly expanding storage capacities. The new chkdsk design makes administrators aware of file system health at all times and provides convenient scheduled resolution with nearly zero downtime. Using this new design, downtime for correcting typical corruptions is less than two seconds. The correction process is now split into the following phases to ensure coordinated, rapid, and transparent corruption resolution. The following figure shows that in the green phases, the volume remains online. The final phase requires the volume to be offline for only seconds (which is indicated in red). In the previous model, the entire resolution process would be red rather than the stripe.
What’s new in NTFS health model
Unlike the previous two-state model, the redesigned NTFS health model implements four health states. Some states are for informational purposes only, whereas some require action. The health states are the following:
■ Online and healthy This state represents normal operation. In this state, there is no user action required and no corruptions are detected.
■ Online spot verification needed The file system is briefly placed in a transient state after it detects a corruption that cannot be self-healed. The volume is placed in this state only until the spot verification service verifies the corruption. This state does not require any user action.
■ Online scan needed After the spot verification service confirms the corruption, the file system is placed in this state. During the next maintenance window, an online scan is automatically performed. If more convenient, the online scan can also be run manually from the Action Center. During the online scan, all verified issues and
corrective actions are logged for later execution.
■ Spot fix needed The file system puts the corrupted volume in this state after the online scan is completed and will notify the user via the Action Center. Corrupted system volumes must be taken offline for corrections. Corruptions of removable media and other non-system volumes can be corrected without taking the system offline by selecting “Error checking” under the Tools tab of the volume’s Properties menu. For Windows Server 2012 systems, spot fix corrections on data volumes can be scheduled during maintenance windows.