RAID, performance tests - Revision history

Jimbo: Reverted edits by 173.88.199.104 (talk) to last revision by Jimbo

2012-08-25T21:55:23Z

Reverted edits by 173.88.199.104 (talk) to last revision by Jimbo

173.88.199.104 at 14:18, 13 August 2012

2012-08-13T14:18:01Z

Jimbo: Reverted edits by 173.57.12.54 (talk) to last revision by Jimbo

2012-03-08T13:15:20Z

Reverted edits by 173.57.12.54 (talk) to last revision by Jimbo

173.57.12.54: /* RAID3/RAID4: Stripe Set with Parity Member */

2012-02-28T16:23:16Z

RAID3/RAID4: Stripe Set with Parity Member

Jimbo: Reverted edits by 70.166.24.207 (talk) to last revision by 71.133.223.218

2012-01-24T21:49:41Z

Reverted edits by 70.166.24.207 (talk) to last revision by 71.133.223.218

70.166.24.207: /* RAID5: Stripe Set with Distributed Parity */

2012-01-16T18:54:11Z

RAID5: Stripe Set with Distributed Parity

71.133.223.218: /* RAID5: Stripe Set with Distributed Parity */ Incorrectly said RAID 5 was extremely similar to RAID 1 except for the parity location -- changed to RAID 3.

2011-11-20T04:08:25Z

RAID5: Stripe Set with Distributed Parity: Incorrectly said RAID 5 was extremely similar to RAID 1 except for the parity location -- changed to RAID 3.

Jimbo at 16:05, 6 March 2011

2011-03-06T16:05:43Z

Jimbo: Reverted edits by 82.231.131.106 (Talk) to last revision by Jimbo

2011-03-06T16:03:06Z

Reverted edits by 82.231.131.106 (Talk) to last revision by Jimbo

82.231.131.106: /* RAID5: Stripe Set with Distributed Parity */

2011-03-06T10:06:22Z

RAID5: Stripe Set with Distributed Parity

← Older revision		Revision as of 21:55, 25 August 2012
Line 115:		Line 115:
	Googling "gmirror performance" and "gmirror slow" did not get me much of a return; just one other individual wondering why his gmirror was so abominably slow - so I reformatted the test system with 6.2-RELEASE (i386) and retested. Unfortunately, the gmirror results did not improve with the change of platform back to i386. It strikes me as very odd that graid3 with only 3 drives (therefore only 2 data drives) outperforms even a ''five''-drive gmirror implementation. And in sharp contrast to '''gmirror''', of course, the Linux kernel RAID1 results speak for themselves.		Googling "gmirror performance" and "gmirror slow" did not get me much of a return; just one other individual wondering why his gmirror was so abominably slow - so I reformatted the test system with 6.2-RELEASE (i386) and retested. Unfortunately, the gmirror results did not improve with the change of platform back to i386. It strikes me as very odd that graid3 with only 3 drives (therefore only 2 data drives) outperforms even a ''five''-drive gmirror implementation. And in sharp contrast to '''gmirror''', of course, the Linux kernel RAID1 results speak for themselves.

−	[[Category: Configuring FreeBSD]]	+	[[Category: Common Tasks]] [[Category: FreeBSD for Servers]] [[Category: FreeBSD for Workstations]] [[Category: Configuring FreeBSD]] [[Category: RAID]]

← Older revision		Revision as of 14:18, 13 August 2012
Line 115:		Line 115:
	Googling "gmirror performance" and "gmirror slow" did not get me much of a return; just one other individual wondering why his gmirror was so abominably slow - so I reformatted the test system with 6.2-RELEASE (i386) and retested. Unfortunately, the gmirror results did not improve with the change of platform back to i386. It strikes me as very odd that graid3 with only 3 drives (therefore only 2 data drives) outperforms even a ''five''-drive gmirror implementation. And in sharp contrast to '''gmirror''', of course, the Linux kernel RAID1 results speak for themselves.		Googling "gmirror performance" and "gmirror slow" did not get me much of a return; just one other individual wondering why his gmirror was so abominably slow - so I reformatted the test system with 6.2-RELEASE (i386) and retested. Unfortunately, the gmirror results did not improve with the change of platform back to i386. It strikes me as very odd that graid3 with only 3 drives (therefore only 2 data drives) outperforms even a ''five''-drive gmirror implementation. And in sharp contrast to '''gmirror''', of course, the Linux kernel RAID1 results speak for themselves.

−	~~[[Category: Common Tasks]] [[Category: FreeBSD for Servers]] [[Category: FreeBSD for Workstations]]~~ [[Category: Configuring FreeBSD~~]] [[Category: RAID~~]]	+	[[Category: Configuring FreeBSD]]

@@ Line 18: / Line 18: @@
 Use of the parity member in read operations is optional with FreeBSD's RAID3 implementation ([[graid3]]), but testing seems to show round-robin parity member reads offering little to no performance benefit with potentially severe performance decreases, so it doesn't seem to be a good idea for most settings.
-'''Graid3''' only allows array sizes of 2^n+1 - ie 3, 5, 9, 17, and so on.  This may be a significant handicap to those trying to squeeze the absolute most out of an array without breaking the bank.  Linux's RAID4 does not suffer from this limitation.
+'''Graid3''' only allows array sizes of 2^n-1 - ie 3, 5, 9, and so on.  This may be a significant handicap to those trying to squeeze the absolute most out of an array without breaking the bank.  Linux's RAID4 does not suffer from this limitation.
 RAID3 is extremely rare outside the FreeBSD world - I have neither seen nor heard of it in actual use other than with '''graid3''' under FreeBSD.  It is similarly rare to see RAID4 outside Linux's implementation.
 == RAID5: Stripe Set with Distributed Parity ==

@@ Line 18: / Line 18: @@
 Use of the parity member in read operations is optional with FreeBSD's RAID3 implementation ([[graid3]]), but testing seems to show round-robin parity member reads offering little to no performance benefit with potentially severe performance decreases, so it doesn't seem to be a good idea for most settings.
-'''Graid3''' only allows array sizes of 2^n-1 - ie 3, 5, 9, and so on.  This may be a significant handicap to those trying to squeeze the absolute most out of an array without breaking the bank.  Linux's RAID4 does not suffer from this limitation.
+'''Graid3''' only allows array sizes of 2^n+1 - ie 3, 5, 9, 17, and so on.  This may be a significant handicap to those trying to squeeze the absolute most out of an array without breaking the bank.  Linux's RAID4 does not suffer from this limitation.
 RAID3 is extremely rare outside the FreeBSD world - I have neither seen nor heard of it in actual use other than with '''graid3''' under FreeBSD.  It is similarly rare to see RAID4 outside Linux's implementation.
 == RAID5: Stripe Set with Distributed Parity ==

@@ Line 24: / Line 24: @@
 == RAID5: Stripe Set with Distributed Parity ==
-RAID5 is extremely similar to RAID3, except that parity blocks are distributed among all member drives - where a RAID3 array with 5 member drives will write 4 blocks to members 1,2,3,4 and then a parity block to member 5, a RAID5 array with 5 member drives will write 4 blocks to members 1,2,3,4 and then a parity block to member 5 - and on the next cycle, will write 4 data blocks to 2,3,4,5 and then a parity block to member 1. Successive writes will eventually distribute the parity block to all member drives, thus avoiding the write bottleneck that occurs in the RAID3 dedicated parity drive.
+RAID5 is extremely similar to RAID3, except that parity blocks are distributed among all member drives - where a RAID3 array with 5 member drives will write 4 blocks to members 1,2,3,4 and then a parity block to member 5, a RAID5 array with 5 member drives will instead write ''five'' blocks to members 1,2,3,4,5 and then a parity block to member 1 - and on the next cycle, will write five data blocks to 2,3,4,5,1 and then a parity block to member 2.
 RAID5 and RAID3 are obviously extremely similar. Although RAID5 has dominated the industry as a whole for decades, there has been some controversy in recent years over their relative performance levels.  We will test both types in this article.

← Older revision		Revision as of 16:05, 6 March 2011
Line 28:		Line 28:
	RAID5 and RAID3 are obviously extremely similar. Although RAID5 has dominated the industry as a whole for decades, there has been some controversy in recent years over their relative performance levels. We will test both types in this article.		RAID5 and RAID3 are obviously extremely similar. Although RAID5 has dominated the industry as a whole for decades, there has been some controversy in recent years over their relative performance levels. We will test both types in this article.

−	FreeBSD does not currently support RAID5 at all. Even inexpensive "hardware" controllers such as the Nvidia onboard types (which use the CPU to XOR data blocks to generate parity and vice versa) are unsupported.	+	FreeBSD does not currently support software RAID5 at all. Even inexpensive "hardware" controllers such as the Nvidia onboard types (which use the CPU to XOR data blocks to generate parity and vice versa, and are frequently referred to as "fakeraid") are unsupported. True hardware controllers which do not use the CPU for parity calculation are, of course, supported.

@@ Line 28: / Line 28: @@
 RAID5 and RAID3 are obviously extremely similar. Although RAID5 has dominated the industry as a whole for decades, there has been some controversy in recent years over their relative performance levels.  We will test both types in this article.
-FreeBSD does not support software RAID5 (only raid 0/1/3; raid5 can be achieved with zfs/raidZ). FreeBSD supports hardware raid5 controllers (3ware, adaptec, Promise, LSI Megaraid, etc...). This exclude cheap disigned controllers (such as the Nvidia that's use the CPU to XOR data blocks to generate parity and vice versa).
+FreeBSD does not currently support RAID5 at all.  Even inexpensive "hardware" controllers such as the Nvidia onboard types (which use the CPU to XOR data blocks to generate parity and vice versa) are unsupported.
-Keep in mind that FreeBSD is mostly used for industrial applications: it's not to be installed on the latest low-cost chinese motherboard, to play X11 and the latest nvidia 3d hardware and proprietary drivers, you have windows or ubuntu for this purpose.
 = Results =