W4118 Operating Systems

1. W4118 Operating Systems Instructor: Junfeng Yang

2. Disk • Goal • Learn how disks work • This knowledge will help us better understand tradeoffs in file systems • Disk Structure • Disk Interface • Disk Overhead • Disk Technology Trends • Redundant Arrays of Inexpensive Disks (RAID)

3. Moving-head Disk Structure

4. Disk Interface • Disk drives addressed as one-dimension of logical sectors • The logical sector is the smallest unit of transfer • This array is mapped sequentially onto disk sectors • Logical sector 0 is the 1st sector of 1st track of the outermost cylinder • Logical sector address incremented within track, then within tracks of cylinder, then across cylinders, from outermost to innermost • Precisely reverse-engineer this mapping in OS is difficult • Disk transparently remap defective sectors • Number of sectors per track is not a constant

5. Disk Overheads • To read from disk, disk controller computes cylinder #, surface #, sector # • Latency includes: • Seek time: to get to the track • Latency time: to get to the sector (rotation delay) • Transfer time: get bits off the disk Track Sector Rotation Delay Seek Time

6. Sequential v.s. Random • Overhead of sequential access • Seek to the right track • Rotate to the right sector • Transfer • Overhead of random access of the same amount of data • Seek to the right track • Rotate to the right sector • Transfer • Repeat • Since seek and rotate are slow, sequential access is much faster than random access

7. Modern Disk Parameters

8. Disk Technology Trends • Data  More dense • More bits per square inch • Disk head closer to surface • Create smaller disk with same capacity • Disk geometry  smaller • Spin faster  Increase bandwidth, reduce rotational delay • Faster seek • Lighter weight • Disk price  cheaper • Density improving more than speed (mechanical limitations)

9. RAID Motivation • Performance • disks are slow compared to CPU • Disk speed improves slowly compared to CPU • Reliability • In single disk systems, one disk failure  data loss • Cost • A single fast, reliable disk is expensive

10. RAID Idea • RAID idea: use redundancy to improve performance and reliability • Redundant array of cheap disks as one storage unit • Fast: simultaneous read and write disks in the array • Reliable: use parity to detect and correct errors • RAID can have different redundancy levels, achieving different performance and reliability • Seven different RAID levels (0-6)

11. Evaluating RAID • Performance • Large Read • Large Write • Large Read-Modify-Write • Small Read • Small Write • Small Read-Modify-Write • Reliability • Tolerance of disk failures • Cost • Storage utilization: data capacity / total capacity

12. RAID 0: Non-redundant Striping • Structure • Data striped across all disks in an array • No parity • Advantages: • Good performance: with N disks, speed up N times • Disadvantages: • Poor reliability: one disk failure  data loss

13. RAID 1: Mirroring • Structure • Keep a mirrored (shadow) copy of data • Advantages • Good reliability: one disk failure OK • Good read performance • Disadvantage • High cost: one data disk requires one parity disk

14. RAID2: Error-Correction Parity parity disks • Structure • A data sector striped across data disks • Compute error-correcting parity and store in parity disks • Advantages • Good reliability with higher storage utilization than mirroring • Disadvantages • Cost still high • Poor small read, write, read-modify-write performance • E.g., to read a data sector, must read a sector from each disk

15. RAID3: Bit-Interleaved Parity • Structure • Each data sector striped across data disks • One parity disk (XOR of each stripe of a data sector) • Advantages • Same reliability with one disk failure as RAID2 since disk controller can determine what disk fails • High storage utilization • Disadvantages • Poor small read, write, read-modify-write performance

16. RAID4: Block-Interleaved Parity P • Structure • A set of data sectors (parity group) striped across data disks • One parity disk (XOR of data sectors) • Advantages • Same reliability as RAID3 since disk controller can detect if sector is correct • Good small read performance • Disadvantages • Poor small write and read-modify-write performance • All writes must write parity  parity disk is bottleneck

17. RAID5: Block-Interleaved Distributed Parity P P P P • Structure • Same as RAID4 except no single parity disk • Parity sectors distributed across all disks • Advantages • Good performance • Good Small write and read-modify-write performance • Disadvantages • Only tolerate one disk failure

18. RAID6: P+Q Redundancy P Q Q P P Q P Q • Structure • Same as RAID 5 except using two parity sectors per parity group • Advantages • Can tolerate two disk failures

19. RAID Levels

20. New Mass Storage Technologies • Disk speed limited by mechanical parts • Seek and rotational delays • New memory-based mass storage technologies solve this problem • NAND Flash • Battery-backed DRAM (NVRAM) • Disadvantages • Price: still more expensive than same capacity disk • Reliability: more likely to lose data