Whiskey Tango Foxtrot

[temporus]

Can anyone explain to me why a disk can be advertised as 750 GB, but when put into the system, I am told that it is a 621 GB drive?

Seriously.  Even if I grant that when they convert up, each time it's by orders ot 1000, instead of 1024 as it should be, and I do the appropriate division, we're still talking a conversion rate that would put me at closer to 698 GB.  Still, that's a pretty huge chunk to lose to formatting, anyway you look at it.

Grrrr.

It's either a loss of 18% of advertised space to the file structure, or a loss of 11%.  Either way, I think that's pretty much BS.  I suspect those disks are not really what they are advertised as, no matter which way you think about it.   Yes, yes, I know that the vendor stressed multiple times that the disk space we were purchasing was based upon the RAW disk.  But you could provide reasonable estimates of what the space we would end up with, instead of just empasizing total raw disk space every time in the discussion.

Bah!

Comments

[edhorch.livejournal.com]

The answer depends on the file system type. Different file system types have differing levels of overhead, and some of it is artificial. Some file systems preallocate journal space for recovery in case of improper shutdown. Others have "snapshot" capabilities, which allow you to freeze the filesystem at some point in time so that it can be recovered in case you do something really bad, and that takes up extra space. There are some file systems whose performance goes WAY down as they get close to full, so they artificially knock 10% off the advertised capacity. Also, any sort of RAID and/or volume management will add some overhead, independent of what's in those volumes.

If this is the primary disk on a new computer, chances are that there's a "recovery" partition on the drive that's not visible to the OS. This is a common cheap substitute for providing OS and application CDs with the machine. Instead, you get a recovery CD, which boots a minimal OS and then reloads the machine from the contents of the recovery partition. (Yes, too bad if the disk dies.)

This is one of the reasons that I have to keep telling my management that disks may be cheap but fast, reliable, accessible, persistent, always-on and flexibly manageable storage isn't.

[temporus.livejournal.com]

This is pre-snapshot. I can set the amount of snapshot I want, and that will reduce it even further. (But that, I udnerstand as an option for me to consider.)

It's just frustrating, because as of this moment, it appears that I'll be getting at BEST between 1/3 and 1/2 of that RAW space as actual, enduser available disk space. I'm sorry, but that's WAY too much overhead in my book.

[edhorch.livejournal.com]

You're right, that is too much overhead. I have a NetApp that has 4TB of raw space (28 146GB drives) and about 2.5TB of usable (pre-snapshot) space after RAID and volume overhead. I also have a SnapServer with 1TB raw (4 250GB drives) and about 650GB of post-RAID usable space.

[dqg-neal.livejournal.com]

I didn't think you were supposed to lose over 9% until after you hit a terabyte. I would have expected closer to the 698GB myself.

Which drive is this one? i wanna make sure I don't buy any.

[temporus.livejournal.com]

According to the service tech, this is industry standard, and has nothing to do with their equipment. He says they market them this way, but the actual accessible space, regardless of manufacturer is 621.

I know that it's true that advertised sizes never match what your system sees. Even before I format my IBM server disks, they'll drop from the advertised 73 GB down to like 68. But that's small potatoes compared to the kind of theft we're talking here.

[dqg-neal.livejournal.com]

The service tech is lying or misinformed. Definintely not every manufacturer with that issue. Every other 750GB drive I've formatted ended up using averaged about 690, roughly 7-8% loss. Under 250 GB it used to be about a 5-6% loss.

[temporus.livejournal.com]

Which is actually even less than that. Because apparently, drive manufacturers insist upon using the 1000 not the 1024 multipliers. They do it on purpose, and since going by units of 1000 is correct via the metric system, it's not apparently lying. Even though in my opinion it is. So that explains a loss of 50 GB off the top. Still, there's a pretty damn big difference between the two.