Jump to content
derek500

OS X Defragmentation tools for disk backup volume? Please share your experiences

Recommended Posts

I am looking for a utility to defragment our main backup volume. We use a D2D2T strategy. We have a Lacie 4BIG attached by eSata to a 2007 Mac Pro, dual 2.66 Xeon 6GB RAM. It is a 6TB volume in a RAID 5 hardware config with ~3TB in use by Retrospect and ~3TB available (retrospect is the only thing using this volume). D2D/regular client backups seem fine and run at normal/network speeds. At first the D2T part ran well, fast speeds and completed steadily, but lately transferring the 3TB to tape/SCSI has slowed from 1-2 days to 3-4 days. It has been in use for 6 months with weekly grooming to keep the most recent 10 backups.

 

I have tested smaller test volumes/backups on firewire drives to tape with Retrospect and the transfer speed seems normal. I think it's time to try defragmenting the backup volume. I did have to rebuild the main catalog file for the disk volume a few weeks ago, and that didn't have any positive or negative affect on my D2T transfer. (I was hoping that would bring a positive change!) I start a new catalog for each tape set every two weeks, and it's just the initial transfer that is dragging things out.

 

What OS X defrag tools have you used, any major or minor pros/cons? I assume "Optimizing" isn't going to have any kind of positive affect on this volume? Any of these have a free trial that actually works that I can check out? (I'm not necessarily looking for a free utility, just don't want to be forced to purchase to try something that may or may not do what I need) Some Unix command line I've forgotten about?

 

How frequently do you find you need to defragment your backup volume? Should I stop the Retrospect engine while defragmenting?

 

Please mention your experiences with the tool as it relates to this purpose, beyond just the name or that it worked great on your macbook pro etc... How long has your backup volume been in use, how frequently do you groom, etc. Also please mention if you have been using/grooming the same backup disk set without defragmenting since v8 came out and don't have any issues transferring...

 

Thanks

-Derek

Share this post


Link to post
Share on other sites

I have never defragmented my Lacie 4Big (exact same one that you have, I believe), but I have defragmented the internal hard disk that contains my OS and my compressed catalog files (which range from 5G to 10G) using iDefrag a couple of times to test things out. There was no noticeable difference in the speed of backups -- or grooming (which is what I was really testing out) -- after the defragmentation at all. iDefrag (which I've used elsewhere) has been perfectly safe for me and I've never had any issues with the program. (Whether or not it actually *does anything useful*, I don't really know...)

 

I occasionally to D2D2D to keep a backup of the Lacie to two smaller (and older) Lacie raid drives using Carbon Copy Cloner. The speeds for doing that backup of the volume seem pretty reasonable to me.

 

I groom 2-3 media sets weekly and have gotten down to about 300G of free space on the drive (I have about 1.5 TB free at the moment) and have been using this particular drive since April 2009. Your post made me look at this and iDefrag says my fragmentation on this drive is 18.6/30.9% (which wouldn't be unexpected considering every disk media set on this volume has been groomed probably at least 50 times -- some of them get groomed weekly, so those have been groomed even more often -- many hundreds of times by now.

 

However, I can't imagine that defragging this volume would make that much of difference in doing my final "2D" step considering how much time it would probably take to optimize 3TB of data and how fragmented the drive would get again after just a few months.

 

 

But if I *were* to do this, I'd certainly do it with the engine off. No question about that...

Share this post


Link to post
Share on other sites

Defragmentation tools have never gained much acceptance on Unix-based systems (including OS X). Disk allocation management tends to work well on these systems until the HDDs get nearly full. One indicator of that full disk problem was that on the old BSD Fast File System (now often just called UFS or Unix File System) didn't allow normal (non-root) users to allocate space on the HDD if it was 95% or more full.

 

I'm not surprised that Maser hasn't noticed any disk performance improvement after defragmnting a HDD.

Share this post


Link to post
Share on other sites

At work I run D2D2T on Retrospect for Windows 7.6. After four years of weekly grooming, the disks became very fragmented. After defragging them as well as the OS (C:) drive, there was NO conceivable difference in the performance before and after the defragging.

Share this post


Link to post
Share on other sites

I occasionally to D2D2D to keep a backup of the Lacie to two smaller (and older) Lacie raid drives using Carbon Copy Cloner. The speeds for doing that backup of the volume seem pretty reasonable to me.

Have you ever used Retrospect to do this backup? Just curious what the performance is like. I don't have another drive of this size to play with, we just transfer to tape instead.

When watching the process (via the "Summary" tab) while it's slow it seems to spend a lot of time comparing the catalog file, as if it's trying to decide what to transfer next.

Is your backup volume/catalog the same since 2009?

At work I run D2D2T on Retrospect for Windows 7.6. After four years of weekly grooming, the disks became very fragmented. After defragging them as well as the OS (C:) drive, there was NO conceivable difference in the performance before and after the defragging.

Thanks. Were you having performance trouble with the D2T part? Did you find anything that made it better?

 

As we groom to keep 10 backups, and have offsite tapes for DR, I would think a worst case scenario of recycling the entire backup set and re-copying everything from the clients might take a few days, but overall really wouldn't be a giant impact. I could think of that as a twice-a-year task if it makes a big enough improvement. Also that would help me purge out some of the really old backups that I have to go seek and destroy manually (retired systems, etc

Share this post


Link to post
Share on other sites

Thanks. Were you having performance trouble with the D2T part? Did you find anything that made it better?

Not really. We see performance go down slightly sometimes for no apparent reason.

A reboot of the Windows' NAS-servers and the Retrospect server every other month seems to keep performance up.

Share this post


Link to post
Share on other sites

Have you ever used Retrospect to do this backup? Just curious what the performance is like. I don't have another drive of this size to play with, we just transfer to tape instead.

When watching the process (via the "Summary" tab) while it's slow it seems to spend a lot of time comparing the catalog file, as if it's trying to decide what to transfer next.

Is your backup volume/catalog the same since 2009?

 

 

 

I use Carbon Copy Cloner to do the final "TD" copying from my backup volume to my backup-backup volumes. For this (incremental backups of backups), it's faster than Retrospect ever would be.

 

My backup volume is the same since 2009, but I have rebuilt most of my catalog files at least once (for testing purposes) at least once -- I think my oldest is about 18 months old, but most are about a year old at this point.

Share this post


Link to post
Share on other sites

Thanks for all the feedback. I started getting errors like this:

!Trouble positioning: "1-Retro Main 6TB Disk" (2154899937), error -102 ( trouble communicating)

 

by the thousands. I tried changing my connection from the SATA/eSATA cable to using FW800 and still got the same errors, so it looks like I have a problem with the Lacie 4BIG unit that needs looking into.

Share this post


Link to post
Share on other sites

Just a final update to this issue. I opened a support case with Retrospect and they confirmed this as a disk/hardware issue. I discussed the problem at length with Lacie and they had me do a low level (zero all data) format on the volume. I had to start my entire backup set over, but since the low level format performance has been steady, and I haven't had one single error from grooming or transfer to tape scripts.

 

I recommended to Retrospect that they add a note to their user guide that whenever setting up a volume for grooming/long term use that a low level format is an important step before starting. I learned it the hard way. Erase with zero all data on any disk volume before using it for backups!

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

×