Lee H. Marzke on 9 Dec 2011 19:30:50 -0800

[Date Prev] [Date Next] [Thread Prev] [Thread Next] [Date Index] [Thread Index]

Re: [PLUG] Lost gigabytes?

> > Date: Thu, 8 Dec 2011 15:47:29 -0500
>  > From: Rich Freeman > <r-plug@thefreemanclan.net>
> <snip>
> > Your one partition is further divided up using LVM, and so root is
> > just a portion of it.  LVM does allow for resizing so you can
> > shuffle
> > things around - IF YOU KNOW WHAT YOU ARE DOING!!!!!!!!!  You may
> > need
> > to shrink existing filesystems and then their corresponding logical
> > volumes - then increase other logical volumes and their
> > corresponding
> > filesystems.

The terminology between normal partitions,  LVM,  and ZFS is a bit

POD                       LVM                      ZFS
xxx                     Physical Volume            vDev
Disk                   Volume Group (VG)          Pool
Fixed Partition      Logical Volume (LV)          Folder

At the lowest level,  a Fixed partition compares to a LV.  You can
resize the LV just like a partition.   With ZFS the comparable item
is the folder - however the folder is sized by quota - so there
there no more manual overhead in resizing volumes.

The partitions or LV's obtain their storage from the overlying Disk
or VG, but they are fixed in size. With ZFS the Folder just allocates
dynamically from the zpool with optional quota.

The VG's obtain space from one or more partitions ( PV's ) which
can span multiple physical disks, while the zpool stripes data across
multiple virtual Devices (vDevs).   vDev's are always
redundant ( either mirrors, or Raid-Z,  or Raid-Z2 ).   A common
configuration is either 2-disk mirrors or 3-disk Raid-Z vDev's with
a total of 6 or more vDev's in the pool.

So with 11 disks in ZFS you don't use 9 data and 1 parity and 1 spare
for RAID-5,  you use 5 (2 disk) mirrors and a spare, or 3 (3-disk)
RaidZ vDev's and 2 spares.  Each ZFS pool can also have optional
SSD read cache disk attached as a level 2 adaptative replacment
cache (L2ARC) , and an optional 2-disk SSD mirror attached for the
ZFS intent log ( ZIL ) which speeds up writing by converting 5 seconds
of random writes into a sequential flush to the main pool.

Both Linux and ZFS use any spare RAM as a read cache,  which ZFS
calls the Level 1 ARC  or  just the (ARC).

To make it even more confusing ( some vendors like Nexenta )
have renamed the ZFS conponents in the GUI to resemble the
more standard terms.

Zpool ->  Volume
Folder -> Share

ZFS has NFS built-in and CIFS services integrated.  So you just 'share' the
folder via NFS, CIFS, by setting properties and enabling the
service, although it does make use Samba for CIFS 
There is no exports file to manage.   There is also
a built in idmap facility to map Windows AD users and groups
to Unix users and the reverse. ( However in practice with windows
security ID's and ACL's it just gets too complex to deal with ).

If you need block storage you create a zvol instead of a folder
and then enable iSCSI and setup a target.  

>JP Vossen, CISSP     
> The other nice thing about LVM is that it allows for "snapshots."
> Maybe.  The capability is there, but to actually use it requires that
> you have some "free extents" which you will not have if you did a
> guided
> setup in Ubuntu.  (I filed a bug about that a long time ago.)
> Basically, if you create a snapshot, the CoW (copy on write) data has
> to
> go somewhere, and if all space is already used, you're toast.

The other issue with LVM is that you can't revert to the snapshot.
So you can only use LVM for helping with freezing data for a backup
and not preserving the system to back out a bad patch.

"Between subtle shading and the absence of light lies the nuance of iqlusion..." - Kryptos 

Lee Marzke, lee@marzke.net http://marzke.net/lee/ 
IT Consultant, VMware, VCenter, SAN storage, infrastructure, SW CM 
+1 800-393-5217 office +1 484-348-2230 fax 
+1 610-564-4932 cell sip://8003935217@4aero.com VOIP 

Philadelphia Linux Users Group         --        http://www.phillylinux.org
Announcements - http://lists.phillylinux.org/mailman/listinfo/plug-announce
General Discussion  --   http://lists.phillylinux.org/mailman/listinfo/plug