[LINK] google services

[jim birch]

On 16/02/07, andrew clarke <mail at ozzmosis.com> wrote:
>
> I imagine they are hoping (probably quite reasonably) that their users
> will only use a small fraction of their allowed storage space, and thus
> only budgeting for that.


I would expect that virtually all of the big content of most people's emails
is simply duplicated material.  Surely Google would hash attachments and
images then store a only one copy (well, a couple or so, as their system has
redundancy.)  They detect quoted text and probably also do something similar
with form letters.  They're also good at wiping junk.  This would cut the
problem down by a couple of orders of magnitude, or so.

Gmail allows you to archive messages indefinitely but don't most users
simply delete.  My email history is bunk, anyway.  There are utilities
around that allow you to use gmail as network storage but this wouldn't have
a big user base.  Google's upper limit of storage would only affect a tiny
proportion of users so it wouldn't be their main concern.  It would be
interesting to see their usage figures but I expect they are trade secrets.

More important for Google, would be the average user storage after
deduplication, redundancy, compression, indexing, etc.  It's a way way
smaller figure so they won't have anything like user_count x
allowable_storage spinning around aimlessly.

I'd also be interested to know what they spend per Gb of hard disk space.
If you can buy cheap retail disks for around AUD0.7 / Gb Google probably pay
under 0.2 for a lesser spec disk that can be swapped out of their redundant
storage system on failure and binned or returned for remanufacture.  Of
course, there's the cost of the stuff required to house and talk to all them
disks and the quarterly power bill must look like an international telephone
number.   But overall, provided the system works, the 3 Gb of "apparent"
storage they offer would be pretty cheap (per user).

Google also have/had a project called GDrive where you buy a chunk of their
storage.  It's been under development for a couple of years and hasn't made
it out of the incubator.  I'm don't know if is technical or commercial or
other.  personally, I'd be happy to pay a moderate sum for a remotely cached
guaranteed storage area that would persist through hardware upgrades, disk
crashes, location changes, etc.  The stuff on my home pc that I really want
to keep is mosty photos, a bit of business stuff and some documents.  I
don't do mp3.  I could probably get by on
<object:"gmail-like_incrementing_counter"/> Gb.

Jim