SSD recommendation

mir

mir

Famous Member
Apr 14, 2012
3,576
136
133
Copenhagen, Denmark
Hi all,

I am about to purchase a new SSD drive. I am considering one of these:
- Corsair force series GT
- OCZ Vertex 3/4
- Intel 330
- Crucial M4

Do any of you have any experience with one of these and have some recommendations?

Thanks in advance.
 
Last edited:
I run all ocz. I have a vertex 2, trouble free for two years, and just put a vertex 3 in my laptop.

Sent from my DROID RAZR using Tapatalk 2
 
I did quick test with vertex 4 128GB firmware 1.3 on SATA2 with defaults on ext4.
Results: cca Buffer reads cca 200MB/s and 57 fsyncs/second.

I also tested crucial m4 64GB.
ext3: 1600fsync/s, BR 220MB/s
ext4 with default: 150fsync/s, BR 220MB/s
ext4 with discard,noatime,barrier=0,nodiratime: cca 200-300fsync/s, BR 220MB/s
ext4 with noatime,barrier=0,nodiratime: 600fsync/s, BR 220MB/s

i'm not sure about mount options, i think you should test with nodelalloc.


edit: barrier should be set 0 not 1.

I got new motherboard (with sata3 support) today:
on ext3: cca 2600fsyncs/s
on ext4 with noop: cca 2800fsyncs/s

Code: 
   root@pve21-1:~# cat /proc/mounts | grep pve-root 
 /dev/mapper/pve-root / ext4 rw,noatime,nodiratime,errors=remount-ro,barrier=0,nodelalloc,data=ordered 0 0

Code: 
root@pve21-1:~# pveperf
CPU BOGOMIPS:      10375.89
REGEX/SECOND:      1145065
HD SIZE:           14.52 GB (/dev/mapper/pve-root)
BUFFERED READS:    200.29 MB/sec
AVERAGE SEEK TIME: 0.07 ms
[B]FSYNCS/SECOND:     2843.73[/B]
DNS EXT:           117.23 ms
DNS INT:           1.08 ms (local.lan)

with discard:

Code: 
   CPU BOGOMIPS:      10375.39 
 REGEX/SECOND:      1170950 
 HD SIZE:           14.52 GB (/dev/mapper/pve-root) 
 BUFFERED READS:    201.07 MB/sec 
 AVERAGE SEEK TIME: 0.08 ms 
 [B]FSYNCS/SECOND:     528.39 [/B]
 DNS EXT:           254.70 ms 
 DNS INT:           1.02 ms (local.lan)


disk: crucial m4 64GB
 
Last edited:
snowman66 said:
I did quick test with vertex 4 128GB firmware 1.3 on SATA2 with defaults on ext4.
Results: cca Buffer reads cca 200MB/s and 57 fsyncs/second.

I also tested crucial m4 64GB.
ext3: 1600fsync/s, BR 220MB/s
ext4 with default: 150fsync/s, BR 220MB/s
ext4 with discard,noatime,barrier=1,nodiratime: cca 200-300fsync/s, BR 220MB/s
ext4 with noatime,barrier=1,nodiratime: 600fsync/s, BR 220MB/s

i'm not sure about mount options, i think you should test with nodelalloc.
Click to expand...
What does fsync/s exactly say? I have a hunch that the difference between ext3 and ext4 for some reason is not actually stating anything else other that the test in some way is given a false picture of the difference. I do not beleave that ext3 should be that much better than ext4.

I have also made tests which shows the dramatical effect of mounting with discard. In my opinion the only answer to this is that the backported implementation of Redhat in some way is severaly broken. Does anybody have some test from a kernel 3.2?

nodelalloc on the promox 2.6.32 is also reducing the numbers so for SSD on proxmox and ext4 avoid discard and nodelalloc which might indicate that one should go for ext3?
 
Last edited:
+1 for the OCZ Vertex 4
The Samsung 830 model is a good alternative, too.

Both have advanced, internal garbage collection, so you will not have to care about native TRIM support in the OS.
 
mir said:
Hi all,

I am about to purchase a new SSD drive. I am considering one of these:
- Corsair force series GT
- OCZ Vertex 3/4
- Intel 330
- Crucial M4

Do any of you have any experience with one of these and have some recommendations?

Thanks in advance.
Click to expand...
Hi,
I have some OCZ Vertex3, but also some trouble with that (ok, was not the best idea to use this kind of SSD as backup-spooldisk).
More stable (and internal garbage collection - so you can use it in raids) is a Corsair Performance Pro.

Udo
 
Code: 
pveperf - the Proxmox benchmark

=head1 SYNOPSIS

pveperf [PATH]

=head1 DESCRIPTION

Tries to gather some CPU/Hardisk performance data on the hardisk
mounted at PATH (/ is used as default) 

* CPU BOGOMIPS: bogomips sum of all CPUs

* REGEX/SECOND: regular expressions per second (perl performance test), should be above 300000

* HD SIZE: harddisk size

* BUFFERED READS: simple HD read test. Modern HDs should reach 
at least 40 MB/sec

* AVERAGE SEEK TIME: tests average seek time. Fast SCSI HDs reach values < 8 milliseconds. Common IDE/SATA disks get values from 15 to 20 ms.

*[B] FSYNCS/SECOND[/B]: value should be greater than 200 (you should 
enable 'write back' cache mode on you RAID controller - needs a battery 
backed cache (BBWC)).

* DNS EXT: average time to resolve an external DNS name

* DNS INT: average time to resolve a local DNS name
 
Ive got Vertex 4 60gb Firmware 1.5 on sata 3

BR 246 mb/s
fsync/s 3957
 
One month of use 98% left.

# smartctl -a /dev/sda
smartctl 5.40 2010-07-12 r3124 [x86_64-unknown-linux-gnu] (local build)
Copyright (C) 2002-10 by Bruce Allen, http://smartmontools.sourceforge.net

=== START OF INFORMATION SECTION ===
Model Family: OCZ Vertex SSD
Device Model: OCZ-VERTEX4
Serial Number: OCZ-0IPCPRN04Z0Z3E61
Firmware Version: 1.5
User Capacity: 128,035,676,160 bytes
Device is: In smartctl database [for details use: -P show]
ATA Version is: 9
ATA Standard is: Exact ATA specification draft version not indicated
Local Time is: Wed Aug 29 07:47:35 2012 UTC
SMART support is: Available - device has SMART capability.
SMART support is: Enabled

=== START OF READ SMART DATA SECTION ===
SMART overall-health self-assessment test result: PASSED


SMART Attributes Data Structure revision number: 18
Vendor Specific SMART Attributes with Thresholds:

[TABLE="width: 500"]
[TR]
[TD]ID#[/TD]
[TD]ATTRIBUTE_NAME[/TD]
[TD]FLAG[/TD]
[TD]VALUE[/TD]
[TD]WORST[/TD]
[TD]THRESH[/TD]
[TD]TYPE[/TD]
[TD]UPDATED[/TD]
[TD]WHEN_FAILED[/TD]
[TD]RAW_VALUE[/TD]
[/TR]
[TR]
[TD]1[/TD]
[TD]Raw_Read_Error_Rate[/TD]
[TD]0x0000[/TD]
[TD="align: center"]---[/TD]
[TD="align: center"]---[/TD]
[TD="align: center"]---[/TD]
[TD]Old_age[/TD]
[TD="align: center"]Offline[/TD]
[TD="align: center"]-[/TD]
[TD]393222[/TD]
[/TR]
[TR]
[TD]3[/TD]
[TD]Spin_Up_Time[/TD]
[TD]0x0000[/TD]
[TD="align: center"]100[/TD]
[TD="align: center"]100[/TD]
[TD="align: center"]000[/TD]
[TD]Old_age[/TD]
[TD="align: center"]Offline[/TD]
[TD="align: center"]-[/TD]
[TD]0[/TD]
[/TR]
[TR]
[TD]4[/TD]
[TD]Start_Stop_Count[/TD]
[TD]0x0000[/TD]
[TD="align: center"]100[/TD]
[TD="align: center"]100[/TD]
[TD="align: center"]000[/TD]
[TD]Old_age[/TD]
[TD="align: center"]Offline[/TD]
[TD="align: center"]-[/TD]
[TD]0[/TD]
[/TR]
[TR]
[TD]5[/TD]
[TD]Reallocated_Sector_Ct[/TD]
[TD]0x0000[/TD]
[TD="align: center"]100[/TD]
[TD="align: center"]100[/TD]
[TD="align: center"]000[/TD]
[TD]Old_age[/TD]
[TD="align: center"]Offline[/TD]
[TD="align: center"]-[/TD]
[TD]0[/TD]
[/TR]
[TR]
[TD]9[/TD]
[TD]Power_On_Hours[/TD]
[TD]0x0000[/TD]
[TD="align: center"]---[/TD]
[TD="align: center"]---[/TD]
[TD="align: center"]---[/TD]
[TD]Old_age[/TD]
[TD="align: center"]Offline[/TD]
[TD="align: center"]-[/TD]
[TD]48325732[/TD]
[/TR]
[TR]
[TD]12[/TD]
[TD]Power_Cycle_Count[/TD]
[TD]0x0000[/TD]
[TD="align: center"]---[/TD]
[TD="align: center"]---[/TD]
[TD="align: center"]---[/TD]
[TD]Old_age[/TD]
[TD="align: center"]Offline[/TD]
[TD="align: center"]-[/TD]
[TD]877668[/TD]
[/TR]
[TR]
[TD]232[/TD]
[TD]Available_Reservd_Space[/TD]
[TD]0x0000[/TD]
[TD="align: center"]100[/TD]
[TD="align: center"]100[/TD]
[TD="align: center"]000[/TD]
[TD]Old_age[/TD]
[TD="align: center"]Offline[/TD]
[TD="align: center"]-[/TD]
[TD]8782065532[/TD]
[/TR]
[TR]
[TD]233[/TD]
[TD]Media_Wearout_Indicator[/TD]
[TD]0x0000[/TD]
[TD="align: center"]098[/TD]
[TD="align: center"]000[/TD]
[TD="align: center"]000[/TD]
[TD]Old_age[/TD]
[TD="align: center"]Offline[/TD]
[TD="align: center"]-[/TD]
[TD]98[/TD]
[/TR]
[/TABLE]
Click to expand...
 
Last edited:
in order to see see the current mounts:
Code: 
cat /proc/mounts

if you want to set other options, add it in:

Code: 
/etc/fstab
 
mir said:
What do you try to emphasize or point out with your copy of the output from smartctl? Maybe I have missed something important:-)
Click to expand...

From wikipedia

233 0xE9 Media Wearout Indicator - Intel SSD reports a normalized value of 100 (when the SSD is new) and declines to a minimum value of 1. It decreases while the NAND erase cycles increase from 0 to the maximum-rated cycles
Click to expand...
 
So your reading indicates that a vertex 4 should be able to survive 50 month ~= 4 years?
OCZ gives 5 years of guaranty so it seems like a bad business decision for them? ;-)
 
mir said:
So your reading indicates that a vertex 4 should be able to survive 50 month ~= 4 years?
OCZ gives 5 years of guaranty so it seems like a bad business decision for them? ;-)
Click to expand...

They give guaranty that device will works but ssd size will become smaller and smaller
 
You must log in or register to reply here.
Top Bottom