1 / 28

I/O and Disk Storage: Performance, Reliability, and Connectivity

This chapter introduces the concepts of I/O devices, disk storage, buses, and performance measures, as well as discussing the goals and constraints of I/O systems. It also covers the anatomy of a disk drive, reliability considerations, and RAID techniques for improving disk performance and reliability. Additionally, it explores the connectivity of I/O devices through buses and synchronization methods.

jforrest
Download Presentation

I/O and Disk Storage: Performance, Reliability, and Connectivity

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. I/O – Chapter 8 • Introduction - 8.1 • Disk Storage and Dependability – 8.2 • Buses and other connectors – 8.4 • I/O performance measures – 8.6

  2. Input / Ouput devices • __________________________________ • keyboard, mouse, printer, game controllers, … • __________________________________ • hard drive, zip drive, … • __________________________________ • music, video, …

  3. Goals & Constraints

  4. Taxonomy • Behavior – • Input (____________) • Output (_______________________) • Storage (______________________________) • Partner • Human or machine on other side? • Data rate (speed) • ______________________________________

  5. Measures of Performance • Response time • __________ – time a user must wait for task • Bandwidth • _________________ per unit time • _________________ per unit time

  6. Anatomy of a Disk Drive

  7. Vocabulary • Head – the device that reads data from a disk • Each disk is divided into ________ _______ called _________ • Each track is made up of _________ sectors • cylinder – volume of all _______ that lie under the heads at a given point on all surfaces • nonvolatile – data that remains even when ______ is removed

  8. Vocabulary • seek – the act of positioning the _____ over the correct ________ • rotational delay or latency – average latency to rotate the ______ over the correct _______ • transfer time – time required to _________ a block of data • disk controller – controls disk accesses

  9. Example 1 - Performance • What is the average time to read or write a 512-byte sector for a typical disk rotating at 10,000 RPM? The advertised average seek time is 6 ms, the transfer rate is 50 MB/sec, and the controller overhead is 0.2 ms. Assume that the disk is idle, so that there is no waiting time.

  10. Reliability • Reliability – measure of a continuously working system • Availability – how often, on average, the system is working properly • MTTF – Mean Time to Failure • MTTR – Mean Time to Repair • MTBF – Mean time between failures

  11. Availability • Availability – how often, on average, the system is working properly • Availability = MTTF / (MTTF + MTTR)

  12. Improving MTTF • Fault avoidance • Fault tolerance • Fault forecasting

  13. RAID - Redundant Arrays of Inexpensive Disks • Shift from one large disk to several small disks • Cheaper, smaller, faster • Inherently less reliable • Provide redundancy to counteract lower reliability

  14. RAID 0 • No redundancy!!! • Only a performance increase • Striping (interleaving) – allocation of logically sequential blocks to separate disks to increase performance • Parallel access controlled by disk controller – computer knows nothing about it.

  15. RAID 1 • mirroring – write the identical data to multiple disks • Requires twice as many disks as RAID 0 • If a disk fails, use the backup copy, move to a working set of mirrored space.

  16. RAID 3 • Bit-interleaved parity • Store only enough data to recover original • Group N blocks • Add one bit of parity – xor of all bits. • Lost data can be reconstructed by looking at the rest of the bits in the group.

  17. On a write • Read all blocks of data in parity group • Calculate new parity • Write new block • Write new parity

  18. RAID 4 • More efficient parity update • On write: • Read old data • xor with new data • adjust parity • Write parity, Write new data

  19. RAID 5 • Rotate parity blocks around system • Spread out writing (since parity always written)

  20. Summary • RAID 1 and RAID 5 most common • 80% of server disks use RAID • Repair: • hot swapping – replace disks with power on • Standby spares – spares included in system for immediate reconstruction of data

  21. Connecting I/O Devices • Much slower than processor / memory • Support lots of heterogeneous devices

  22. Bus • control lines – send / receive commands • data lines – transfer data • processor-memory bus – fast, small bus connecting DRAM to processor • I/O bus – slow, long bus connecting many devices to system through a controller.

  23. Synchrony • ___________________ – clock in control line, fixed protocol is relative to clock. • _______________________ – no clock – must coordinate through hand-shaking to determine when data is ready to send / receive.

  24. hand-shaking • Series of steps used to coordinate bus transfers. Both parties must acknowledge they are ready before moving to next step. • Control lines: • ReadyReq: proc/device wants to read • DataRdy: dev/proc is ready to send data • Ack: acknowledge ReadReq or DataRdy

  25. Metric Units • Memory: GB = 2^30 • I/O: GB = 10^9 • Be careful when reading specs • For this class, we will pretend that all use base-2 units. • Throughput more important than latency • Large database operations (TP – Transaction Processing)

  26. Example 1 • Execution time = 100 seconds • 90 seconds CPU time, 10 sec I/O time • CPU time improves by 50%/yr for 5 years • I/O does not improve • How much faster is program after 5 years? • What percentage of new time is I/O?

  27. Example 2 • System A: • .005 sec per I/O op • 4 overlapping I/O ops at a time • System B: • .002 sec per I/O op • no overlapping I/O ops • Which has the higher throughput?

More Related