Solid State Drives

From ArchWiki
Revision as of 20:44, 30 June 2010 by Graysky (Talk | contribs) (Created page with '=High Level Summary of Article= As most Archers know, Solid State Drives (SSDs) are not PnP devices. Special considerations such as partition alignment, choice of file system, T…')

(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to: navigation, search

High Level Summary of Article

As most Archers know, Solid State Drives (SSDs) are not PnP devices. Special considerations such as partition alignment, choice of file system, TRIM support, etc. are needed to setup SSDs for optimal performance. This article attempts to capture referenced, key learnings to enable users to get the most out of SSDs under Arch (Linux in general).

  • Use a "modern" SSD with native TRIM support
  • Use kernel version >=2.6.33 to make use of kernel-based TRIM
  • Align SSD partitions to the SSD's unique EBS (erase block size)
  • Use the correct mount flags (noatime,discard for ext4)
  • Select an appropriate IO Scheduler (noop or deadline for SSDs/cfq for HDDs)
  • Reseting an SSD to a Virgin State (optional)
Note: Readers are encouraged to contribute to enhance the quality of this article.

Pre-Purchase Considerations

There are several key features to look for prior to purchasing a contemporary SSD.

Key Features

  • Native TRIM support is a vital feature that both prolongs SSD lifetime and reduces the loss of performance over time.
  • Buying the right size SSD is also key. As with all your filesystems, target <75 % occupancy for all SSD partitions to ensure efficient use by the kernel.

On-line Reviews

This section is not meant to be all-inclusive, but does capture some key reviews.

Partitions

System Partition Scheme

An overarching theme for SSD usage 'simplicity' in terms of locating high-read/write partitions on a physical HDD rather than on an SSD. Doing so will add life to an SSD. For example, consider relocating the /var partition to a physical disc on the system rather than on the SSD itself to avoid read/write wear. Many users elect to keep only /boot, /, and /home on the SSD locating /var and /tmp on a physical HDD - or better yet, into Random Access Memory (RAM) provided the system has enough to spare. See the next section for more on this procedure.

Locate /tmp in RAM

For systems with >=4 gigs of memory, locating /tmp in the RAM is desirable and easily achieved by first clearing the physical /tmp partition and then mounting it to tmpfs (RAM) in the Template:Filename. The following line gives an example:

none	/tmp	tmpfs	nodev,nosuid,nodiratime,noatime,size=1000M,mode=1777	0	0

Locate Browser Profiles in RAM

For the same reason outlined above, one can easily mount one's firefox profile (and others such as chromium, etc.) into RAM via tmpfs. For more on this procedure, see the Speed-up Firefox Using tmpfs article. In addition to the obvious speed enhancements, users will also save read/write cycles on their SSD by doing so.

Physical Partitioning and Alignment

Proper partition alignment is key for optimal performance and longevity. The community seems to be in agreement on the use of fdisk as the utility of choice for partitioning SSDs (although one can also find guides whose authors advocate using parted). Consensus opinion on the best settings for number of heads and cylinders is tough to find. There seem to be two different camps on this issue as shown below. The key is that one needs to align the partitions based on the SSDs EBS (erase block size). The Intel X25-M for example uses an EBS of 512 KiB.

Ted Tso recommends using a setting of 224/56 for SSDs with an EBS of 512 KiB:

# fdisk -H 224 -S 56 /dev/sdX

While others advocate a setting of 32/32 for SSDs with an EBS of 512 KiB:

# fdisk -H 32 -S 32 /dev/sdX

Additional Reading

File Systems

Many options exist for file systems including ext2, ext3, ext4, XFS, and btrfs. Initially, ext2 was thought to be a good choice as it lacks journaling which would avoid extraneous read/write cycles. Ext4 can also be used without a journal and is thought to be superior to ext2 in a number of areas. The obvious drawback of using a non-journaling file system is data loss as a result of an ungraceful dismount (i.e. post power failure). With modern SSDs, Ted Tso advocates that journaling can be enabled with minimal extraneous read/write cycles under most circumstances:

Amount of data written (in megabytes) on an ext4 file system mounted with noatime.

operation journal w/o journal percent change
git clone 367.0 353.0 3.81 %
make 207.6 199.4 3.95 %
make clean 6.45 3.73 42.17 %

"What the results show is that metadata-heavy workloads, such as make clean, do result in almost twice the amount data written to disk. This is to be expected, since all changes to metadata blocks are first written to the journal and the journal transaction committed before the metadata is written to their final location on disk. However, for more common workloads where we are writing data as well as modifying filesystem metadata blocks, the difference is much smaller."

Btrfs support has been included with the mainline 2.6.29 release of the Linux kernel. Some feel that it is not mature enough for production use while there are also early adopters of this potential successor to ext4. It should be noted that at the time this article was written (27-June-2010), a stable version of btrfs does not exist. See this blog entry for more on btrfs.

Mount Flags in /etc/fstab

There are several key mount flags to use in one's Template:Filename entries for SSD partitions.

  • noatime - Reading accesses to the file system will no longer result in an update to the atime information associated with the file. The importance of the noatime setting is that it eliminates the need by the system to make writes to the file system for files which are simply being read. Since writes can be somewhat expensive, this can result in measurable performance gains. Note that the write time information to a file will continue to be updated anytime the file is written to with this option enabled.
  • discard - The discard flag will enable the benefits of the TRIM command so long as one is using kernel version >=2.6.33.
/dev/sda1 / ext4 defaults,noatime,discard 0 1
Warning: Users need to be certain that kernel version 2.6.33 or above is being used AND that their SSD supports TRIM before attempting to mount a partition with the discard flag. Data loss can occur!

I/O Scheduler

Consider switching from the cfq scheduler to the noop or deadline scheduler. Using the noop scheduler for example simplifies requests in the order they are received, without giving any consideration to where the data physically resides on the disk. This option is thought to be advantageous SSDs since seek times are identical for all sectors on the SSD. For more on schedulers, see this Linux-mag article.

The cfq scheduler is enabled by default on Arch. Verify this by viewing the contents /sys/block/sda/queue/scheduler:

$ cat /sys/block/sdX/queue/scheduler
noop deadline [cfq]

The scheduler currently in use is denoted from the available schedulers by the brackets. To switch to another scheduler (noop in this example), one can add the following line in Template:Filename:

# echo noop > /sys/block/sdX/queue/scheduler
Note: Only switch the scheduler to noop or deadline for SSDs. Keeping the cfq scheduler for all other physical HDDs is highly recommended.

Swap Space on SSDs

One can place a swap partition on an SSD. Note that most modern desktops with an excess of 2 Gigs of memory rarely use swap at all. The notable exception is systems which make use of the hibernate feature. The following is recommended tweak for SSDs using a swap partition that will reduce the "swapiness" of the system thus avoiding writes to swap.

# echo 1 > /proc/sys/vm/swapiness

Or one can simply modify Template:Filename as recommended in the Maximizing Performance wiki article.

vm.swappiness=20
vm.vfs_cache_pressure=50

Reseting an SSD to a Virgin State

One can use a technique to literally reset an SSD to its factory virgin state marking all its cells as empty thus restoring it to its factory default write performance.

Warning: Doing so will erase ALL data on the SSD and rendering it unrecoverable by even data recovery services!

To successfully issue an ATA Security Erase command, one need to first set a user password. This step is omitted from almost all other sources which describe how to secure erase with hdparm.

The example output shown is from an INTEL X25-M G1 80GB SSD running 8820 firmware. It was run from an Ubuntu 9.04 32-bit (Jaunty) Live CD booted from a USB flash drive.

Step 1 - Make sure the drive Security is not frozen:

Issue the following command:

# hdparm -I /dev/sdX

If the command output shows "frozen" one cannot continue to the next step. Most BIOSes block (do no allow) the ATA Secure Erase command by issuing a "SECURITY FREEZE" command to "freeze" the drive before booting an operating system.

A possible solution for SATA drives is hot-(re)plug the data cable (which might crash the kernel). If hot-(re)pluging the SATA data cable crashes the kernel try letting the operating system fully boot up, then quickly hot-(re)plug both the SATA power and data cables.

  • It has been reported that hooking up the drive to an eSATA SIIG ExpressCard/54 with an eSATA enclosure will leave the drive security state to "not frozen".
  • Placing the target system into "sleep" (Clevo M865TU notebook) has reported to work as well; this may reset other drives to "not frozen."
Security: 
        Master password revision code = 65534
                supported
        not     enabled
        not     locked
        not     frozen
        not     expired: security count
                supported: enhanced erase
        2min for SECURITY ERASE UNIT. 2min for ENHANCED SECURITY ERASE UNIT.

Step 2 - Enable security by setting a user password

Note: When the user password is set the drive will be locked after next power cycle denying normal access until unlocked with the correct password).

Any password will do, as this should only be temporary. After the secure erase the password will be set back to NULL. In this example, the password is "Eins" as shown:

# hdparm --user-master u --security-set-pass Eins /dev/sdX
security_password="Eins"
/dev/sdX:
Issuing SECURITY_SET_PASS command, password="Eins", user=user, mode=high

As a sanity check, issue the following command

# hdparm -I /dev/sdX

The command output should display "enabled":

 Security: 
        Master password revision code = 65534
                supported
                enabled
        not     locked
        not     frozen
        not     expired: security count
                supported: enhanced erase
        Security level high
        2min for SECURITY ERASE UNIT. 2min for ENHANCED SECURITY ERASE UNIT.

Step 3 - Issue the ATA Secure Erase command:

# time hdparm --user-master u --security-erase Eins /dev/sdX

Wait until the command completes. This example output shows it took about 40 seconds for an Intel X25-M 80GB SSD, for a 1TB hard disk it might take 3 hours or more!

security_password="Eins"
/dev/sdX:
Issuing SECURITY_ERASE command, password="Eins", user=user
0.000u 0.000s 0:39.71 0.0%      0+0k 0+0io 0pf+0w

The drive is now erased. After a successful erasure the drive security should automatically be set to disabled (thus no longer requiring a password for access). Verify this by running the following command:

# hdparm -I /dev/sdX

The command output should display "not enabled":

 Security: 
        Master password revision code = 65534
                supported
        not     enabled
        not     locked
        not     frozen
        not     expired: security count
                supported: enhanced erase
        2min for SECURITY ERASE UNIT. 2min for ENHANCED SECURITY ERASE UNIT.