GoConqr has a new look! Let us know what you think.

Sign up free

Flashcards

Chapter 12

Ali Raza Bhangu Ch12 DB Uni English 65 cards

Cards in this deck (65)

  1. 1Volatile StorageLoses contents when power off
  2. 2non-volatile storagecontents persist when power is off
    includes secondary, tertiary storage as well as batter backed up main memory
  3. 3Factors Affecting Choice of Storage MediaSpeed with which data can be accessed
    Cost per unit of data
    reliability
  4. 4primary storagefastest media but volatile (cache, main memory)
  5. 5Secondary Storagenext level in hierarchy, non-volatile, moderately fast access time
    Also called on-line storage
    ex: flash memory, magnetic disks
  6. 6tertiary storagelowest level in hierarchy, non-volatile, slow access time
    Called off-line storage, used for archival storage, ex: magnetic tape, optical storage
  7. 7Magnetic TapeSequential access, 1 - 12 TB capacity
    A few drives with many tapes
    Juke boxes with petabytes (1000's of TB) of storage
  8. 8SATA (Serial ATA)SATA 3 supports data transfer of speeds up to 6gb/s
  9. 9SAS (Serial Attached SCSI)SAS V3 supports 12gb/sec
  10. 10NVMe (Non-volatile Memory Express) InterfaceWorks with PCIe connectors to support lower latency and higher transfer rates
    Supports data transfer of 24gigs/sec
  11. 11Storage Area Networks (SAN)A large number of disks are connected by a high-speed network to a number of servers
  12. 12Network Attached Storage (NAS)provides a file system interface using networked file system protocol, instead of providing a disk system interface
  13. 13Read-Write HeadSurface of platter divided into circular tracks
    Over 50k-100k tracks per platter on typical hard disks
  14. 14Sectorsthe smallest unit of data that can be written or read
    Size typically 512 bytes
    Typical Sectors per track: 500 - 1000 on inner tracks, 1000 - 2000 on outer tracks
  15. 15How to read/write a sector?Disk arm swings to position head on track
    Platter spins continually data is R/W as sector passes under head.
  16. 16Head-Disk AsembeliesMultiple disk platters on a single spindle 1 - 5 usually
    One head per platter, mounted on common arm
  17. 17Disk ControllerInterfaces b/w the computer system and the disk hardware
    Accepts high-lvl commands to read or write a sector
    Initiates actions such as moving the disk arm to right track and reading or writing the data
    Computers and attaches checksums to each sector to ensure data is read back correctly,
    If data is corrupted, stored checksum wont match recomputed checksum
    Ensures succesful of writing by reading back sector after writing it
    Performs remapping of bad sectors
  18. 18Access TimeTime it takes from when a read or write is requested to when data transfer begins
    Consists of Seek Time, Rotational Latency
  19. 19Seek TimeTime it takes to reposition arm over correct track
    Avg Seek time is 1/2 worst case seek time
    Would be 1/3 if all tracks had same number of sectors and we ignore the start/stop arm movement time
    4 - 10 milliseconds on typical disk
  20. 20Rotational latencyTime it takes for sector to be access to appear under the head
    4 - 11 milliseconds on avg disk (5400 - 15000 RPM)
    Average latency is 1/2 of above latency
    Overall latency is 5 - 20 Msec depending on disk model
  21. 21Data Transfer RateRate at which data can be retrieved or stored to disk
    25 - 200 MB per second max rate, lower for inner tracks
  22. 22Disk Blocklogical unit for storage allocation and retrieval
    4 - 16 Kilobytes typically
    Smaller Blocks: More transfers from disk
    Larger Blocks: More space wasted due to partially filled blocks
  23. 23Sequential Access PatternSuccessive requests are for successive disk blocks
    Disk seek required only for first block
  24. 24Random Access PatternSuccesive request for blocks that can be anywhere on disk
    Each access requries a seek
    Low transfer rates due to seeks
  25. 25I/O Per Second (IOPS)Number of random block reads that a disk can support per second
    50 - 200 IOPS current gen magentic disks
  26. 26Mean Time to Failure (MTTF)Avg time disk is expected to run without any failure.
    Typically 3 - 5 Years
    Decreases as disk ages
  27. 27NAND FlashWidely used for storage, cheaper than NOR
    Requires Page at a time read ( page: 512 bytes - 4KB)
    Not much difference b/w sequential and random read
    Page can only be written once, must be erased to allow rewrite
  28. 28Solid State Disks SSDStandard block oriented disk interfaces, but store data on multiple flash storage devices internally
    Transfer rate of up to 500Mb/Sec using SATA
    3GB/Sec using NVMe PCIe
  29. 29How does erase happen?Happens in units of erase block
    Takes 2-5 milliseconds
    Typically 256Kb - 1MB (128 Pages - 256)
  30. 30How to avoid waiting for erase?Remapping logical addresses to physical page addresses avoids waiting for erase
  31. 31Flash translation tabletracks mapping, also stored in a label field of flash page
    remapping carried out by flash translation later
  32. 32Wear Levelingafter 100k - 1 million erases, erase block becomes unreliable and cannot be used.
  33. 33Random Reads/Writes Per SecondTypical 4KB Reads: 10k / sec (10k IOPS)
    Typical 4KB Write: 40k IOPS
  34. 34SSD's Parrallel ReadsTypical 4KB Read: 100k IOPS with 32 Requests in parallel on SATA
    350K IOPS with QD-32 on NVMe PCIe
    Typical 4KB Writes:
    100k IOPS with QB32, even higher on some models
  35. 35Data Transfer Rate for Sequential Reads/Writes400MB/Sec for SATA3
    2-3GB/Sec for NVMe PCIe
  36. 36Hybrid DiskCombine small amount of flash cache with larger magnetic disk
  37. 37RAID: Redundant Arrays of Independent DisksDisk org technique that manges large numbers of disks providing aview of a single disk
    - High capacity and high speed by using multiple disks in parralel
    - High reliability by storing data redunatly, so data can be recovered if disk fails.
  38. 38Mirroring or Shadowing ExampleDuplicate every disk. Logical disk consists of two physical disks.
    • Every write is carried out on both disks
    - Reads can take place from either disk
    • If one disk in a pair fails, data still available in the other
    - Data loss would occur only if a disk fails, and its mirror disk also
    fails before the system is repaired
    • Probability of combined event is very small
    -Except for dependent failure modes such as fire or building
    collapse or electrical power surges
  39. 39Mean time to data lossDepends on MTTF and MTTR (time to repair)
  40. 40Main goals of parallelism in disk system?Load balance multiple small accesses to increase throughput
    Parallelize large accesses to reduce response time
    Improve transfer rate by striping data across disks
  41. 41Bit Level StrippingSplit bits of each byte across multiple disks
    - In array of eight disks, write bit i of each byte to disk i
    Each access can read data at 8x rate of single disk
    Seek/Access time is worse than for single disk
    This is not used much anymore
  42. 42Block Level Stripping– with n disks, block i of a file goes to disk (i mod n)
    + 1
    • Requests for different blocks can run in parallel if the blocks reside on
    different disks
    • A request for a long sequence of blocks can utilize all disks in parallel
  43. 43Raid Level 0Block Striping, Non Redundant
    Used in high performance apps where data loss is not critical
  44. 44Raid Level 1Mirrored disks with block stripping
    Offers best write performance
    Popular for apps such as storing log files in a DB system
  45. 45Parity BlocksParity block j stores XOR of bits from block j of each disk
    • When writing data to a block j, parity block j must also be computed
    and written to disk
     Can be done by using old parity block, old value of current block
    and new value of current block (2 block reads + 2 block writes)
     Or by recomputing the parity value using the new values of blocks
    corresponding to the parity block
    • More efficient for writing large amounts of data sequentially
    • To recover data for a block, compute XOR of bits from all other
    blocks in the set including the parity block
  46. 46Raid Level 5Block interleaved Distributed Parity: Partitions data and Partity among all N+1 Disks rather than storing data in N disks and Parity in 1 disk
    - Block writes occur in parallel if the blocks and their parity blocks on different disks
  47. 47Raid Level 6P+Q Redundancy Scheme
    Similar to 5 but stores two error correct blocks (P,Q) instead of single parity block to guard against multiple disk failues
    - Better reliability than Lvl 5 at higher cost
    - Becoming more important as storage size increases
  48. 48Raid Level 2:Not used in Practice
    Memory Style Error Correcting Codes with Bit Striping
  49. 49Raid Level 3Bit Interleaved Parity
  50. 50Raid Level 4Block interleaved parity: uses block level striping and keeps a parity block on a separate parity disk for corresponding blocks from N other disks.
    Raid 5 > Raid 4, since R4 w/h random writes, parity disk gets much higher write load than other disks and becomes a bottleneck
  51. 51Factors for RAID LevelMonetary Cost
    Perfomance: # of IOPS and Bandwith during normal operation
    Performance during failure
    Performance During rebuild of failed disk, including time taken to rebuild failed disk.
  52. 52When is Raid 0 Used?When data safety is not important, data can be recovered quickly from elsewhere
  53. 53When to choose Level 1?Provides much better write performance than Lvl 5
    As Lvl 5 requries at least 2 block reads and 2 block writes to write 1 block, Lvl 1 only needs 2 block writes.
    Lvl 1 Had Higher Storage Cost than Lvl 5
    Prefered for apps with Small / Random Updates
  54. 54When Raid Level 5?Preferred for applications where writes are sequential and large (many blocks), and need large amounts of data storage
  55. 55When to choose Level 6?Gives better data protection than raid 5 since it can tolerate two disk or disk block failures
    Increasing in importance since latent block failures on 1 disk, coupled with failure of another disk can cause data loss in RAID 1 and 5
  56. 56Software RaidRAID implementations done entirely in software, with
    no special hardware support
  57. 57Hardware Raid:RAID implementations with special hardware
    • Use non-volatile RAM to record writes that are being executed
    • Beware: power failure during write can result in corrupted disk
    E.g., failure after writing one block but before writing the second
    in a mirrored system
    - Such corrupted data must be detected when power is restored
    • Recovery from corruption is similar to recovery from failed disk
    • NV-RAM helps to efficiently detected potentially corrupted blocks
    Otherwise all blocks of disk must be read and compared
    with mirror/parity block
  58. 58Latent Failures:data successfully written earlier gets damaged
    • can result in data loss even if only one disk fails
    Hardware Issue
  59. 59Data Scrubbing:• continually scan for latent failures, and recover from copy/parity
  60. 60Hot Swappingreplacement of disk while system is running, without power
    down
    • Supported by some hardware RAID systems,
    • reduces time to recovery, and improves availability greatly
  61. 61How do Hardware Raid Systems ensure that single point of failure will not stop the system?Redundant power supplies with battery backup
    • Multiple controllers and multiple interconnections to guard against
    controller/interconnection failures
  62. 62BufferingIn memory buffer to cache disk blocks
  63. 63Read AheadRead extra blocks from a track in anticipation that they will be requested soon
  64. 64Disk Arm SchedulingAlgorithms reorder block requests so that disk arm movement is minimized.
    Like elevator algorithm.
  65. 65File Organization for Optimization of Disk Block AccessAllocate blocks of a file in as contiguous a manner as possible
    • Allocation in units of extents
    • Files may get fragmented
    E.g., if free blocks on disk are scattered, and newly created file has its blocks scattered over the disk
    Sequential access to a fragmented file results in increased disk
    arm movement
    Some systems have utilities to defragment the file system, in
    order to speed up file access
    - Non-volatile write buffers

Information

Description
Uni Ch12 DB Flashcards on Chapter 12, created by Ali Raza Bhangu on 17/04/2023.
Created by
Ali Raza Bhangu · Apr 2023
Updated
Apr 17, 2023
Views
9
Rating

Feedback

What is it about?

What do you think of the new look?

How do you like it?