We recently were explained the differences between each RAID as well as the difference between thin and thick client computing as well as what Hierarchical Storage Management is.
I am very disappointed that this class is almost over because I have had such a great time learning about all the different types of computers and different software and hard drives that are available.
Created By Shannon DuQuesnay, Copyright 2011
RAID 1 – Mirroring uses a pair of drives
each block of data is written to both drives
block – basic unit of data read/written to storage
single drive failure à no data is
lost, system still runs at full speed, replace failed drive and re-mirror
50% overhead – cost per GB is doubled
“Hot spare” disk – replaces dead disk drive
RAID 0 – Striping 2 or more drives
no extra copies-blocks of data spread across all drives
multiple disks simultaneously working on both read and write requests
the more disks, the faster RAID 0 works
no redundancy à single drive failure à entire RAID 0
fails, all data lost, crashes, replace drive, rebuild array, reinstall OS and
restore data
less reliable than single disk drive
cost per GB is unaffected
RAID 5 3 or more drives
data blocks spread across all drives except one
parity – calculated and then store on remaining drive – used to protect data
parity = 1 if sum of data bits is odd
parity = 0 if sum of data bits is even
fast reads, moderate write speed (must compute parity)
single drive failureà no data lost,
still running but reduced speed à replace failed drive and allow
system to reconstruct the data on the new drive before another failure
hot spare disk – replaces immediately
(N-1) drivers usable for data
common in network servers
RAID 6 any form of raid than can continue to execute in the presence of 2 concurrent
disk failures
capacity = (N-2) disks
fast reads, writes slower, needs hardware support
1TB+ disk HSM: Hierarchical Storage Mgmtsome data used regularly
some data rarely touched
automated, rules-based system allows us to take rarely used from expensive
storage to cheap (unknown to user)
Benefits: expensive high speed storage used for small “working set” and rarely
used data stored to SATA disk or tape
Thick clienthardware – standard Desktop PC
powerful CPU and Lots of RAM
NIC
storage: hard drive, optical, USB
IO: screen, keyboard, mouse, sounds
software: OS, utilities and applications
I am very disappointed that this class is almost over because I have had such a great time learning about all the different types of computers and different software and hard drives that are available.
each block of data is written to both drives
block – basic unit of data read/written to storage
single drive failure à no data is
lost, system still runs at full speed, replace failed drive and re-mirror
50% overhead – cost per GB is doubled
“Hot spare” disk – replaces dead disk drive
no extra copies-blocks of data spread across all drives
multiple disks simultaneously working on both read and write requests
the more disks, the faster RAID 0 works
no redundancy à single drive failure à entire RAID 0
fails, all data lost, crashes, replace drive, rebuild array, reinstall OS and
restore data
less reliable than single disk drive
cost per GB is unaffected
data blocks spread across all drives except one
parity – calculated and then store on remaining drive – used to protect data
parity = 1 if sum of data bits is odd
parity = 0 if sum of data bits is even
fast reads, moderate write speed (must compute parity)
single drive failureà no data lost,
still running but reduced speed à replace failed drive and allow
system to reconstruct the data on the new drive before another failure
hot spare disk – replaces immediately
(N-1) drivers usable for data
common in network servers
disk failures
capacity = (N-2) disks
fast reads, writes slower, needs hardware support
1TB+ disk
some data rarely touched
automated, rules-based system allows us to take rarely used from expensive
storage to cheap (unknown to user)
Benefits: expensive high speed storage used for small “working set” and rarely
used data stored to SATA disk or tape
Thin Client Computing
Thick clienthardware – standard Desktop PCpowerful CPU and Lots of RAM
NIC
storage: hard drive, optical, USB
IO: screen, keyboard, mouse, sounds
software: OS, utilities and applications
Thin Clienthardware – specialized device, very small
modest CPU and RAM
NIC
No Internal hard drive, might have optical or USB
IO: screen, keyboard, mouse , sound
like a terminal: processing and storage occur elsewhere
modest CPU and RAM
NIC
No Internal hard drive, might have optical or USB
IO: screen, keyboard, mouse , sound
like a terminal: processing and storage occur elsewhere
Created by Shannon DuQuesnay, Copyright2011
Actioncentralized CPU, RAM, Storage
use terminal server approach: server OS, terminal server application shares the
servers resources and installed applications among multiple users
application issues – licensing, different versions
virtualization: each user has own VM with OS and applications à standard, unmodified
OS and apps; more flexible, resources are pooled
VM features – deploy from templates, snapshots
use terminal server approach: server OS, terminal server application shares the
servers resources and installed applications among multiple users
application issues – licensing, different versions
virtualization: each user has own VM with OS and applications à standard, unmodified
OS and apps; more flexible, resources are pooled
VM features – deploy from templates, snapshots
Why Thin?more efficient use of HW resources
typical end user PC is underutilized: CPU idling, much unused RAM, vast unused
disk space
centralized CPU, RAM, storage more fully utilized
dynamic allocation of CPU and RAM to users as needed
allocate only necessary amount of storage
allocate resources in nonstandard increments
improved management/support
hardware upgrades: nonstandard increments, less inconvenience to user
centralized backup: data, entire machine (image backup, snapshot)
linked clones: rapid deployment, disk space savings, fast patching, software
updates, pooled computer (retain personality but still fresh daily)
typical end user PC is underutilized: CPU idling, much unused RAM, vast unused
disk space
centralized CPU, RAM, storage more fully utilized
dynamic allocation of CPU and RAM to users as needed
allocate only necessary amount of storage
allocate resources in nonstandard increments
improved management/support
hardware upgrades: nonstandard increments, less inconvenience to user
centralized backup: data, entire machine (image backup, snapshot)
linked clones: rapid deployment, disk space savings, fast patching, software
updates, pooled computer (retain personality but still fresh daily)
No comments:
Post a Comment