TLUG Talk: Storage Technologies and Theory - Revision history

Digimer: /* Level 15 and 51 */

2015-01-02T23:58:31Z

Level 15 and 51

← Older revision		Revision as of 23:58, 2 January 2015
Line 465:		Line 465:
	=== Level 15 and 51 ===		=== Level 15 and 51 ===

	It doesn’t get any more reliable or inefficient than this! RAID 15/51 is the most reliable method of storing data in a single server, combining the raw availability of RAID1 mirrors with the performance and reliability benefits of RAID5. Realistically if this level of reliability was required you would be interested in high availability servers where everything, not just the disk array was redundant.		{{note\|1=After this was written, RAID level 6, and it's combined derivatives '16' and '61' have become popular. They, obviously, are even more reliable.}}

			It doesn’t get any more reliable or inefficient than this! RAID 15/51 is the most reliable method of storing data in a single server, combining the raw availability of RAID1 mirrors with the performance and reliability benefits of RAID5. Realistically if this level of reliability was required you would be interested in high availability servers where everything, not just the disk array was redundant.

	= References =		= References =

Digimer at 13:14, 12 October 2009

2009-10-12T13:14:38Z

← Older revision		Revision as of 13:14, 12 October 2009
Line 1:		Line 1:
	{{talk_header}}		{{talk_header}}

	'''Note''': This talk is quite old, and much has happened in storage tech since this talk was given. I present it here though for historical reasons. Take ~~everything here~~ with a healthy grain a salt.		'''Note''': This talk is quite old, and much has happened in storage tech since this talk was given. This was the first IT-related public talk I gave and as such, it's not my favourite work. I present it here though for historical reasons. Take it all with a healthy grain a salt.

	'''''Storage Technologies and Theory'''''<br />		'''''Storage Technologies and Theory'''''<br />

Digimer at 13:09, 12 October 2009

2009-10-12T13:09:51Z

Show changes

Digimer at 13:08, 12 October 2009

2009-10-12T13:08:40Z

← Older revision		Revision as of 13:08, 12 October 2009
Line 3:		Line 3:
	'''Note''': This talk is quite old, and much has happened in storage tech since this talk was given. I present it here though for historical reasons. Take everything here with a healthy grain a salt.		'''Note''': This talk is quite old, and much has happened in storage tech since this talk was given. I present it here though for historical reasons. Take everything here with a healthy grain a salt.

	'''''Storage Technologies and Theory'''''		'''''Storage Technologies and Theory'''''<br />
	'''An AN!Speech to TLUG, May 2002'''		'''An AN!Speech to TLUG, May 2002'''

Digimer: /* Level 0 */

2009-10-12T13:07:58Z

Level 0

Show changes

Digimer: /* Notes */

2009-10-12T13:04:08Z

Notes

← Older revision		Revision as of 13:04, 12 October 2009
Line 350:		Line 350:
	\|-		\|-
	!style="border-bottom: 1px solid #7f7f7f; text-align: left;"\|Capacity/Spindle RPM		!style="border-bottom: 1px solid #7f7f7f; text-align: left;"\|Capacity/Spindle RPM
	\|120GB / 7,200rpm		\|style="border-bottom: 1px solid #7f7f7f;" class="code"\|120GB / 7,200rpm
	\|73GB / 10,020rpm		\|style="border-bottom: 1px solid #7f7f7f;" class="code"\|73GB / 10,020rpm
	\|36GB / 15,000rpm		\|style="border-bottom: 1px solid #7f7f7f;" class="code"\|36GB / 15,000rpm
	\|-		\|-
	!style="border-bottom: 1px solid #7f7f7f; text-align: left;"\|MSRP in CND / $/GB		!style="border-bottom: 1px solid #7f7f7f; text-align: left;"\|MSRP in CND / $/GB
	\|$572.52 / $4.77		\|style="border-bottom: 1px solid #7f7f7f;" class="code"\|$572.52 / $4.77
	\|$1,518.79 / $20.81		\|style="border-bottom: 1px solid #7f7f7f;" class="code"\|$1,518.79 / $20.81
	\|$1,097.56 / $30.49		\|style="border-bottom: 1px solid #7f7f7f;" class="code"\|$1,097.56 / $30.49
	\|-		\|-
	!style="border-bottom: 1px solid #7f7f7f; text-align: left;"\|Devices per Channel		!style="border-bottom: 1px solid #7f7f7f; text-align: left;"\|Devices per Channel
	\|2		\|style="border-bottom: 1px solid #7f7f7f;" class="code"\|2
	\|15		\|style="border-bottom: 1px solid #7f7f7f;" class="code"\|15
	\|15		\|style="border-bottom: 1px solid #7f7f7f;" class="code"\|15
	\|-		\|-
	!style="border-bottom: 1px solid #7f7f7f; text-align: left;"\|Bus maximum bandwidth		!style="border-bottom: 1px solid #7f7f7f; text-align: left;"\|Bus maximum bandwidth
	\|100MB/Sec		\|style="border-bottom: 1px solid #7f7f7f;" class="code"\|100MB/Sec
	\|160MB/Sec		\|style="border-bottom: 1px solid #7f7f7f;" class="code"\|160MB/Sec
	\|160-320MB/Sec		\|style="border-bottom: 1px solid #7f7f7f;" class="code"\|160-320MB/Sec
	\|-		\|-
	!style="border-bottom: 1px solid #7f7f7f; text-align: left;"\|Data Buffer		!style="border-bottom: 1px solid #7f7f7f; text-align: left;"\|Data Buffer
	\|2,048KB		\|style="border-bottom: 1px solid #7f7f7f;" class="code"\|2,048KB
	\|4,096KB		\|style="border-bottom: 1px solid #7f7f7f;" class="code"\|4,096KB
	\|4,096KB		\|style="border-bottom: 1px solid #7f7f7f;" class="code"\|4,096KB
	\|-		\|-
	!style="border-bottom: 1px solid #7f7f7f; text-align: left;"\|Sustained Data Rate		!style="border-bottom: 1px solid #7f7f7f; text-align: left;"\|Sustained Data Rate
	\|23MB/Sec to 48MB/Sec		\|style="border-bottom: 1px solid #7f7f7f;" class="code"\|23MB/Sec to 48MB/Sec
	\|29MB/Sec to 57MB/Sec		\|style="border-bottom: 1px solid #7f7f7f;" class="code"\|29MB/Sec to 57MB/Sec
	\|36MB/Sec to 53MB/Sec		\|style="border-bottom: 1px solid #7f7f7f;" class="code"\|36MB/Sec to 53MB/Sec
	\|-		\|-
	!style="border-bottom: 1px solid #7f7f7f; text-align: left;"\|Average Seek Time		!style="border-bottom: 1px solid #7f7f7f; text-align: left;"\|Average Seek Time
	\|8.5ms		\|style="border-bottom: 1px solid #7f7f7f;" class="code"\|8.5ms
	\|3.0ms		\|style="border-bottom: 1px solid #7f7f7f;" class="code"\|3.0ms
	\|2.0ms		\|style="border-bottom: 1px solid #7f7f7f;" class="code"\|2.0ms
	\|-		\|-
	!style="border-bottom: 1px solid #7f7f7f; text-align: left;"\|Error Rate		!style="border-bottom: 1px solid #7f7f7f; text-align: left;"\|Error Rate
	\|1 in 10^13bits<br />		\|style="border-bottom: 1px solid #7f7f7f;" class="code"\|1 in 10^13bits<br />
	(1 in 1.2TB)		(1 in 1.2TB)
	\|1 in 10^14bits<br />		\|style="border-bottom: 1px solid #7f7f7f;" class="code"\|1 in 10^14bits<br />
	(1 in 11.3TB)		(1 in 11.3TB)
	\|1 in 10^16bits<br />		\|style="border-bottom: 1px solid #7f7f7f;" class="code"\|1 in 10^16bits<br />
	(1 in 1,136.9TB)		(1 in 1,136.9TB)
	\|-		\|-
	!style="border-bottom: 1px solid #7f7f7f; text-align: left;"\|Rated Start/Stops		!style="border-bottom: 1px solid #7f7f7f; text-align: left;"\|Rated Start/Stops
	\|40,000		\|style="border-bottom: 1px solid #7f7f7f;" class="code"\|40,000
	\|50,000		\|style="border-bottom: 1px solid #7f7f7f;" class="code"\|50,000
	\|50,000		\|style="border-bottom: 1px solid #7f7f7f;" class="code"\|50,000
	\|-		\|-
	!style="~~border-bottom: 1px solid #7f7f7f;~~ text-align: left;"\|Power-on hours/month		!style="text-align: left;"\|Power-on hours/month
	\|333h (11h/day)		\|class="code"\|333h (11h/day)
	\|732h (24h/day)		\|class="code"\|732h (24h/day)
	\|732h (24h/day)		\|class="code"\|732h (24h/day)
	\|}		\|}

Digimer: /* Notes */

2009-10-12T13:02:17Z

Notes

← Older revision		Revision as of 13:02, 12 October 2009
Line 342:		Line 342:

	{\|style="border: 1px solid #7f7f7f;"		{\|style="border: 1px solid #7f7f7f;"
	!colspan="4"\|Table Comparing IDE (120 GXP) vs. SCSI (73LZX) vs. SCSI (36Z15)		!style="border-bottom: 1px solid #7f7f7f;" colspan="4"\|Table Comparing IDE (120 GXP) vs. SCSI (73LZX) vs. SCSI (36Z15)
	\|-		\|-
	\|		\|style="border-bottom: 1px solid #7f7f7f;"\|
	!IBM 120GXP (IDE)		!style="border-bottom: 1px solid #7f7f7f;"\|IBM 120GXP (IDE)
	!IBM 73LZX (SCSI)		!style="border-bottom: 1px solid #7f7f7f;"\|IBM 73LZX (SCSI)
	!IBM 36Z15 (SCSI)		!style="border-bottom: 1px solid #7f7f7f;"\|IBM 36Z15 (SCSI)
	\|-		\|-
	!style="text-align: left;"\|Capacity/Spindle RPM		!style="border-bottom: 1px solid #7f7f7f; text-align: left;"\|Capacity/Spindle RPM
	\|120GB / 7,200rpm		\|120GB / 7,200rpm
	\|73GB / 10,020rpm		\|73GB / 10,020rpm
	\|36GB / 15,000rpm		\|36GB / 15,000rpm
	\|-		\|-
	!style="text-align: left;"\|MSRP in CND / $/GB		!style="border-bottom: 1px solid #7f7f7f; text-align: left;"\|MSRP in CND / $/GB
	\|$572.52 / $4.77		\|$572.52 / $4.77
	\|$1,518.79 / $20.81		\|$1,518.79 / $20.81
	\|$1,097.56 / $30.49		\|$1,097.56 / $30.49
	\|-		\|-
	!style="text-align: left;"\|Devices per Channel		!style="border-bottom: 1px solid #7f7f7f; text-align: left;"\|Devices per Channel
	\|2		\|2
	\|15		\|15
	\|15		\|15
	\|-		\|-
	!style="text-align: left;"\|Bus maximum bandwidth		!style="border-bottom: 1px solid #7f7f7f; text-align: left;"\|Bus maximum bandwidth
	\|100MB/Sec		\|100MB/Sec
	\|160MB/Sec		\|160MB/Sec
	\|160-320MB/Sec		\|160-320MB/Sec
	\|-		\|-
	!style="text-align: left;"\|Data Buffer		!style="border-bottom: 1px solid #7f7f7f; text-align: left;"\|Data Buffer
	\|2,048KB		\|2,048KB
	\|4,096KB		\|4,096KB
	\|4,096KB		\|4,096KB
	\|-		\|-
	!style="text-align: left;"\|Sustained Data Rate		!style="border-bottom: 1px solid #7f7f7f; text-align: left;"\|Sustained Data Rate
	\|23MB/Sec to 48MB/Sec		\|23MB/Sec to 48MB/Sec
	\|29MB/Sec to 57MB/Sec		\|29MB/Sec to 57MB/Sec
	\|36MB/Sec to 53MB/Sec		\|36MB/Sec to 53MB/Sec
	\|-		\|-
	!style="text-align: left;"\|Average Seek Time		!style="border-bottom: 1px solid #7f7f7f; text-align: left;"\|Average Seek Time
	\|8.5ms		\|8.5ms
	\|3.0ms		\|3.0ms
	\|2.0ms		\|2.0ms
	\|-		\|-
	!style="text-align: left;"\|Error Rate		!style="border-bottom: 1px solid #7f7f7f; text-align: left;"\|Error Rate
	\|1 in 10^13bits<br />		\|1 in 10^13bits<br />
	(1 in 1.2TB)		(1 in 1.2TB)
Line 392:		Line 392:
	(1 in 1,136.9TB)		(1 in 1,136.9TB)
	\|-		\|-
	!style="text-align: left;"\|Rated Start/Stops		!style="border-bottom: 1px solid #7f7f7f; text-align: left;"\|Rated Start/Stops
	\|40,000		\|40,000
	\|50,000		\|50,000
	\|50,000		\|50,000
	\|-		\|-
	!style="text-align: left;"\|Power-on hours/month		!style="border-bottom: 1px solid #7f7f7f; text-align: left;"\|Power-on hours/month
	\|333h (11h/day)		\|333h (11h/day)
	\|732h (24h/day)		\|732h (24h/day)
Line 403:		Line 403:
	\|}		\|}

	RAID Levels:		= RAID Levels =

	Level 0 = Stripping, not a “true” RAID level because there is no redundancy. RAID0 is designed to increase performance by cutting data into uniform sized blocks and sequentially writing them to the disks in the array. The total capacity (T) is measured by multiplying the numbers of disks in the array (x) by the storage capacity of the smallest drive(s). T=x*s		== Level 0 ==

			Stripping, not a “true” RAID level because there is no redundancy. RAID0 is designed to increase performance by cutting data into uniform sized blocks and sequentially writing them to the disks in the array. The total capacity (T) is measured by multiplying the numbers of disks in the array (x) by the storage capacity of the smallest drive(s). T=x*s

	Level 1 = Mirroring, the oldest type of RAID. Very popular to it’s ease of use and high-reliability. Write times can be slightly diminished because of the need to write the data across twice the number of disks but a good hardware RAID controller often negates this. Read times are quicker; almost double the speed because the data needs to only be read from one drive. The performance hit during a failure is minimal and the processing power needed to implement RAID1 is small. The main drawback is the initial cost to configure because of the inefficient use of disk space. RAID1 requires an even number of drives. The total storage capacity (T) is measured by multiplying the number of drives in the array (x) by the capacity of the smallest drive (s) and dividing by 2. T=(x*s)/2		Level 1 = Mirroring, the oldest type of RAID. Very popular to it’s ease of use and high-reliability. Write times can be slightly diminished because of the need to write the data across twice the number of disks but a good hardware RAID controller often negates this. Read times are quicker; almost double the speed because the data needs to only be read from one drive. The performance hit during a failure is minimal and the processing power needed to implement RAID1 is small. The main drawback is the initial cost to configure because of the inefficient use of disk space. RAID1 requires an even number of drives. The total storage capacity (T) is measured by multiplying the number of drives in the array (x) by the capacity of the smallest drive (s) and dividing by 2. T=(x*s)/2

Digimer: /* SCSI */

2009-10-12T12:58:07Z

SCSI

Show changes

Digimer: /* Marketing IDE */

2009-10-12T12:49:23Z

Marketing IDE

← Older revision		Revision as of 12:49, 12 October 2009
Line 235:		Line 235:
	The last thing I want to cover regarding IDE is how the speed and performance of a hard drive is advertised. The advertisements often proudly state the speed on the bus that their drive is compatible with. This is usually done in a way that makes most people think that’s the speed they’ll move data at when in fact it’s usually not even close.		The last thing I want to cover regarding IDE is how the speed and performance of a hard drive is advertised. The advertisements often proudly state the speed on the bus that their drive is compatible with. This is usually done in a way that makes most people think that’s the speed they’ll move data at when in fact it’s usually not even close.

	[[Image:tlug_talk_storage_01.png~~\|thumb\|158px~~\|right\|Maxtor "Fast Drives" Logo]]		[[Image:tlug_talk_storage_01.png\|right\|Maxtor "Fast Drives" Logo]]
	Let me beat up on Maxtor because I find they use this confusion to their benefit the most. They proudly state their new “Fast Drives ATA/133” (which isn’t even a ratified standard yet) with the logo on the right.		Let me beat up on Maxtor because I find they use this confusion to their benefit the most. They proudly state their new “Fast Drives ATA/133” (which isn’t even a ratified standard yet) with the logo on the right.

Line 244:		Line 244:
	The Maxtor “DiamondMax/Plus D740X” line, one of Maxtor’s UDMA/133 “Fast Drives”, presents these specs:		The Maxtor “DiamondMax/Plus D740X” line, one of Maxtor’s UDMA/133 “Fast Drives”, presents these specs:

			Fig 2, Specifications of the Maxtor DiamondMax/Plus D740X
			[[Image:tlug_talk_storage_02.png\|center\|Maxtor DiamondMax/Plus D740X specs]]

			http://www.maxtor.com/products/DiamondMax/DiamondMaxPlus/DataSheet/D740X_datasheet.pdf (link is long dead, here: [https://alteeve.com/local/talks/D740X_datasheet.pdf is a local copy] (PDF, 46KiB)



	~~Fig x.x, Specifications of the Maxtor DiamondMax/Plus D740X~~









	~~http://www.maxtor.com/products/DiamondMax/DiamondMaxPlus/DataSheet/D740X_datasheet.pdf~~
	Compare those specs to a “standard” UDMA/100 Maxtor DiamondMax/Plus 60 specs:		Compare those specs to a “standard” UDMA/100 Maxtor DiamondMax/Plus 60 specs:

	Fig ~~x.x~~, Specifications of the Maxtor DiamondMax/Plus 60		Fig 3, Specifications of the Maxtor DiamondMax/Plus 60
			[[Image:tlug_talk_storage_03.png\|center\|Maxtor DiamondMax/Plus 60 specs]]

			http://www.maxtor.com/products/diamondmax/diamondmaxplus/FullSpecs/dmp60spec.pdf (link is long dead, here: [https://alteeve.com/local/talks/dmp60spec.pdf is a local copy] (PDF, 27KiB)

			Notice two things; first off the DiamondMax/Plus 60 doesn’t list the drives transfer rates on the inner and outer tracks of the platter. Secondly though, and more telling is the average transfer rate from the “Media” (drive platters). Assuming Maxtor didn’t change the way they test their own drives it would seem the “slower” DiamondMax/Plus 60 is actually faster than the “Fast Drive” DiamondMax/Plus D740X.

	~~http~~:~~//www~~.~~maxtor.com/products/diamondmax/diamondmaxplus/FullSpecs/dmp60spec.pdf~~		Fig 4, Maxtor’s Stated Burst Rates
			[[Image:tlug_talk_storage_04.png\|center\|Maxtor’s Stated Burst Rates]]

	~~Notice two things; first off~~ the ~~DiamondMax/Plus 60 doesn’t list~~ the ~~drives transfer rates~~ on the ~~inner~~ and ~~outer tracks of~~ the platter. ~~Secondly though,~~ and ~~more telling is~~ the ~~average transfer rate~~ from the ~~“Media” (drive platters). Assuming Maxtor didn’t change~~ the ~~way they test their own drives~~ it would ~~seem~~ the ~~“slower” DiamondMax/Plus 60 is actually faster than~~ the ~~“Fast Drive” DiamondMax/Plus D740X~~.		Let’s look at something else: Burst Rates. This is another great opportunity for marketing people to inflate the performance of their drives. Modern hard drives have dedicated memory (RAM) on the logic board that the drive uses to store cached information. The controller on the drive tries to guess what information will be requested shortly and fetches it from the platter and stores it in cache. If it guessed right then you have a “cache-hit” and the data can be pulled from the drive’s cache and passed to the interface MUCH faster then it would have otherwise taken to get the data off the drive’s platters.

	~~Fig x~~.~~x, Maxtor’s stated Burst Rates~~		http://www.maxtor.com/Maxtorhome.htm (Dead link, now: http://www.maxtor.com/en/index.html)


	~~Let’s look at something else: Burst Rates~~. ~~This is another great opportunity for marketing people to inflate the performance of their drives~~. ~~Modern hard drives have dedicated memory~~ (RAM) on the logic board that the drive uses to store cached information. The controller on the drive tries to guess what information will be requested shortly and fetches it from the platter and stores it in cache. If it guessed right then you have a “cache-hit” and the data can be pulled from the drive’s cache and passed to the interface MUCH faster then it would have otherwise taken to get the data off the drive’s platters.
	http://www.maxtor.com/~~Maxtorhome~~.~~htm~~

	Let me state up front, Cache is a good thing and the more cache the better because the more data you can store in cache the more likely you are to have a cache-hit. That said, cache it not a good way to measure overall performance. Much more often than not the cache doesn’t contain the data needed and the drive must fetch data from the platters anyway negating any benefit the cache may provide. This stated, Maxtor decides again to emphasis the performance of their UDMA/133 drives by quoting maximum transfer rates based on burst rates.		Let me state up front, Cache is a good thing and the more cache the better because the more data you can store in cache the more likely you are to have a cache-hit. That said, cache it not a good way to measure overall performance. Much more often than not the cache doesn’t contain the data needed and the drive must fetch data from the platters anyway negating any benefit the cache may provide. This stated, Maxtor decides again to emphasis the performance of their UDMA/133 drives by quoting maximum transfer rates based on burst rates.
Line 285:		Line 275:
	Look at table x.x at the end of this document for a side-by-side comparison of modern hard drives. These drives represent each company’s best IDE drives as of the time I wrote this.		Look at table x.x at the end of this document for a side-by-side comparison of modern hard drives. These drives represent each company’s best IDE drives as of the time I wrote this.

			= SCSI =
	SCSI:

	Let’s look at “SCSI” now! First up, the performance and marketing considerations I just spoke about regarding IDE also apply to SCSI!		Let’s look at “SCSI” now! First up, the performance and marketing considerations I just spoke about regarding IDE also apply to SCSI!

Digimer: /* Choices Now! */

2009-10-12T12:35:30Z

Choices Now!

← Older revision		Revision as of 12:35, 12 October 2009
Line 185:		Line 185:
	So now we’ve covered all the fundamentals, and you already know what you want to store right? Is a gigabyte a gigabyte a gigabyte? Are all hard drives created equal? Just how are the different RAID levels, well, different? How does fiber and copper stack up?		So now we’ve covered all the fundamentals, and you already know what you want to store right? Is a gigabyte a gigabyte a gigabyte? Are all hard drives created equal? Just how are the different RAID levels, well, different? How does fiber and copper stack up?

			== Foundation ==
	Foundation:

	Hard drives come in two flavours these days called IDE and SCSI. IDE has one interface type and two types of cables. SCSI drives have many interfaces and cables and can be configured in two ways.		Hard drives come in two flavours these days called IDE and SCSI. IDE has one interface type and two types of cables. SCSI drives have many interfaces and cables and can be configured in two ways.

	IDE:		== IDE ==

	First off lets look at IDE, which stands for “Integrated Drive Electronics”. This is by far the most common type of hard drive you will see. In the earlier days of hard drives the controlled for a hard drive was a separate add in card that supported only the drive it was designed for. The name came about when this controller became “integrated” with the physical disk itself, original naming don’t you think?		First off lets look at IDE, which stands for “Integrated Drive Electronics”. This is by far the most common type of hard drive you will see. In the earlier days of hard drives the controlled for a hard drive was a separate add in card that supported only the drive it was designed for. The name came about when this controller became “integrated” with the physical disk itself, original naming don’t you think?
Line 204:		Line 203:
	This is also why older OS’ couldn’t see a device using ATAPI until special drivers where loaded. As time has gone on motherboard BIOS’ have become more intelligent and can now even boot off these devices like the CD-ROM. Eventually ATAPI was integrated into the ATA-4 standard.		This is also why older OS’ couldn’t see a device using ATAPI until special drivers where loaded. As time has gone on motherboard BIOS’ have become more intelligent and can now even boot off these devices like the CD-ROM. Eventually ATAPI was integrated into the ATA-4 standard.

			=== Configuring IDE ===

	~~Configuring~~ IDE:		For those who’ve opened a computer you’ve noticed that IDE devices have three main jumper setting; Master, Slave and Cable Select. The ATA standard allows for two devices per IDE channel meaning then that we can have two devices plus the controller all connected on the same cable. Obviously we needed some way to tell which device we were sending data to or need data from.

	~~For those who’ve opened a computer you’ve noticed that IDE devices have three main jumper setting;~~ Master~~, Slave~~ and ~~Cable Select~~. The ~~ATA standard allows for two devices per IDE channel meaning~~ then ~~that we can have two devices plus~~ the ~~controller all connected~~ on the ~~same~~ cable. ~~Obviously we needed some way~~ to ~~tell which~~ device ~~we were sending data to or need data from.~~		The most common way to do this is to set the first device (or only device) as Master and connect it to the end of the ribbon cable. The second device would then be set via it’s jumper as Slave and cabled to the middle connector on the ribbon cable. Is the position of the cable connectors relative to the setting of the IDE device important? Technically yes but with older 40-pin/40-wire ribbon cables you could usually get away with any configuration

	The most common way to do this is to set the first device (or only device) as Master and connect it to the end of the ribbon cable. The second device would then be set via it’s jumper as Slave and cabled to the middle connector on the ribbon cable. Is the position of the cable connectors relative to the setting of the IDE device important? Technically yes but with older 40-pin/40-wire ribbon cables you could usually get away with any configuration		If you’ve built or worked with a relatively modern computer lately you inevitably noticed the snazzy new cable that came with either your UDMA/33+ hard drive or motherboard. Why the new cable? What was wrong with the old one? The original 40-pin/40-wire ribbon cable was fine for the original slower speeds of older hard drives but the ribbon cable is not terminated and this causes signal bounce and cross-talk that can corrupt data. I’ll talk about signal bounce, cross-talk and termination in better detail shortly when we get into SCSI.

	If you’ve built or worked with a relatively modern computer lately you inevitably noticed the snazzy new cable that came with either your UDMA/33+ hard drive or motherboard. Why the new cable? What was wrong with the old one? The original 40-pin/40-wire ribbon cable was fine for the original slower speeds of older hard drives but the ribbon cable is not terminated and this causes signal bounce and cross-talk that can corrupt data. I’ll talk about signal bounce, cross-talk and termination in better detail shortly when we get into SCSI.

	As modern hard drives reach higher and higher speeds the signal bounce and cross-talk found in the older 40-pin/40-wire cable became too great to ignore. Something had too be done to lessen the distortion while still keeping costs down. From this need the 40-pin/80-wire ATA cable was born! This cable is physically compatible with the older style 40-pin IDC connectors you’ve all seen but is much better at carrying the higher speeds.		As modern hard drives reach higher and higher speeds the signal bounce and cross-talk found in the older 40-pin/40-wire cable became too great to ignore. Something had too be done to lessen the distortion while still keeping costs down. From this need the 40-pin/80-wire ATA cable was born! This cable is physically compatible with the older style 40-pin IDC connectors you’ve all seen but is much better at carrying the higher speeds.
Line 233:		Line 231:
	One more note on IDE cables, both the traditional and 80-conductor ATA IDE cables have a maximum cable length of 45 cm! There are many manufacturers who make longer “IDE” cables but these cables are defiantly not recommended. In the best scenario you will suffer a performance hit, though quite likely you will see data corruptions or unexplained system crashes.		One more note on IDE cables, both the traditional and 80-conductor ATA IDE cables have a maximum cable length of 45 cm! There are many manufacturers who make longer “IDE” cables but these cables are defiantly not recommended. In the best scenario you will suffer a performance hit, though quite likely you will see data corruptions or unexplained system crashes.

			=== Marketing IDE ===

	~~Marketing IDE:~~		The last thing I want to cover regarding IDE is how the speed and performance of a hard drive is advertised. The advertisements often proudly state the speed on the bus that their drive is compatible with. This is usually done in a way that makes most people think that’s the speed they’ll move data at when in fact it’s usually not even close.

	The last thing I want to cover regarding IDE is how the speed and performance of a hard drive is advertised. The advertisements often proudly state the speed on the bus that their drive is compatible with. This is usually done in a way that makes most people think that’s the speed they’ll move data at when in fact it’s usually not even close.

	Let me beat up on Maxtor because I find they use this confusion to their benefit the most. They proudly state their new “Fast Drives ATA/133” (which isn’t even a ratified standard yet) with the logo on the right.		[[Image:tlug_talk_storage_01.png\|thumb\|158px\|right\|Maxtor "Fast Drives" Logo]]
			Let me beat up on Maxtor because I find they use this confusion to their benefit the most. They proudly state their new “Fast Drives ATA/133” (which isn’t even a ratified standard yet) with the logo on the right.

	First up, it’s not the drive that’s actually fast; it’s the interface that is. What you need to look for is either the sustained average external transfer rate or the sustained external transfer rate at the inside and outside edges of the hard drive’s platters.		First up, it’s not the drive that’s actually fast; it’s the interface that is. What you need to look for is either the sustained average external transfer rate or the sustained external transfer rate at the inside and outside edges of the hard drive’s platters.

	Let me again compare using real world examples:		Let me again compare using real world examples:

	The Maxtor “DiamondMax/Plus D740X” line, one of Maxtor’s UDMA/133 “Fast Drives”, presents these specs:		The Maxtor “DiamondMax/Plus D740X” line, one of Maxtor’s UDMA/133 “Fast Drives”, presents these specs: