Read-after-Write Verification
A means of ensuring that data on the network is safe.
NetWare protects data primarily by maintaining duplicate file directories and by redirecting data from bad blocks to reliable blocks on the NetWare server's hard disk.
A hard disk's DET and FAT contain address information that tells the operating system where data can be stored or retrieved.
If the blocks containing these tables are damaged, some or all of the data may be irretrievable.
NetWare greatly reduces the possibility of losing this information by maintaining duplicate copies of the DET and FAT on separate areas of the hard disk.
If one of the blocks in the original tables is damaged, the operating system switches to the duplicate tables to get the location data it needs.
The faulty sector is then listed in the disk's bad block table, and the data it contained is stored elsewhere on the disk.
Every time the server is turned on, the operating system performs a consistency check on both sets of DETs and FATs to verify that the two copies are identical.
If both sets don't match, a warning is sent, and the network supervisor should run VREPAIR.
NetWare hard disks store data in 4, 8, 16, 32, or 64KB blocks. These blocks are specific data storage locations on the disk's magnetic surface. (Block size is the same on all segments of a volume. See also Block.)
Due to the constant reading and writing of data to disk, some storage blocks lose their capacity to store data.
NetWare prevents data from being written to unreliable blocks by employing two complementary features known as read-after-write verification and Hot FixTM .
These features, illustrated in the next two figures, enable a hard disk to maintain the same data integrity it had when it was first tested and installed.
Read-after-write Verification. When data is written to disk, the data is immediately read back from the disk and compared to the original data still in memory.
Figure 11
Read-after-Write Verification
If the data on the disk matches the data in memory, the write operation is considered successful, the data in memory is released, and the next disk I/O operation takes place.
If the data on the disk doesn't match the data in memory, the operating system determines (after making appropriate retries) that the disk storage block is defective.
The Hot Fix feature redirects the original block of data (still in memory) to the Hot Fix Redirection Area, where the data can be stored correctly.
Hot Fix Redirection Area. A small portion of the disk's storage space is set aside as the Hot Fix Redirection Area. This area holds data blocks that are redirected there from faulty blocks on the disk.
Hot Fix is always active unless the disk fails and is inoperative, or the redirection area is full.
Once the operating system records the address of the defective block in a section of the Hot Fix area reserved for that purpose, the server won't attempt to store data in the defective block.
Figure 12
Hot Fix
Read-after-write verification and Hot Fix are transparent. The network supervisor or a console operator can view Hot Fix activity in SERVMAN or MONITOR.
Disk mirroring or duplexing. You can also protect your data with disk mirroring or duplexing.
Mirroring stores the same data on separate disks on the same controller channel; duplexing stores the same data on separate disks on separate controller channels.
Duplexing is the preferred method since two channels rarely fail simultaneously.
See also Directory Entry Table ; Disk mirroring ; Disk duplexing ; System Fault Tolerance.