|
Home > Archive > Data Storage > October 2005 > Computing cost per GB in a SAN environment
You are viewing an archived Text-only version of the thread.
To view this thread in it's original format and/or if you want to reply to
this thread please [click here]
| Author |
Computing cost per GB in a SAN environment
|
|
| dwhitlow1@wi.rr.com 2005-10-13, 5:55 pm |
| Hi All,
What is the best practice for computing the cost per GB within a SAN?
If only computing the cost per GB on a single IDE HDD, it's relatively
simple. However, in a SAN environment, there is so much additional
infrastructure including switches, management software, arrays filled
with varying degrees of capacity, etc. Not to mention that fact that
the array I buy today may not cost the same as the exact same array I
buy tomorrow.
Also, is there some standard for factoring the human capital involved
as well?
I'm doing a study internally for our company and looking for the best
way to do this. The end result is to show the difference between
varying levels of storage including FC vs. iSCSI-based storage, FC vs
SATA HDDs, and things like that.
Any tips on how others have calculated this would be greatly
appreciated.
Thanks,
Don
| |
| _firstname_@lr_dot_los-gatos_dot_ca.us 2005-10-13, 5:55 pm |
| In article <1129217009.150591.47270@o13g2000cwo.googlegroups.com>,
<dwhitlow1@wi.rr.com> wrote:
>What is the best practice for computing the cost per GB within a SAN?
....
>Also, is there some standard for factoring the human capital involved
>as well?
Standard: I don't know. Typical rule of thumb is that operating costs
(floorspace, power, cooling, excluding personnel, maintenance
contracts) at least double purchase cost, often triple it.
Administring a storage system (the personnel cost) is another huge
factor on top of that. I would guess that administration cost could
easily be 3x or 5x the purchase cost.
>I'm doing a study internally for our company and looking for the best
>way to do this.
This is VERY VERY hard to do. If you have the resources of a big
company, order yourself some research reports from Gartner or IDC, and
use them as a starting point only.
> The end result is to show the difference between
>varying levels of storage including FC vs. iSCSI-based storage, FC vs
>SATA HDDs, and things like that.
Everyone wants to do that. Every vendor of product X will try to
prove that product X will greatly reduce the "overhead" costs. It is
difficult or impossible to get unbiased data, unless you happen to
work for one of the large storage vendors, and happen to be the person
who is making up these dishonest claims.
>Any tips on how others have calculated this would be greatly
>appreciated.
I would start with Gartner and IDC, and then talk to people in similar
industries with similar-size data centers. Don't forget to include
costs like the network infrastructure (often charged to a different
department), or backup tape media. And I would take any information
that comes from an interested party (i.e., a vendor) which a huge
grain of salt. If you are talking to vendors, play them off against
each other; they typically have white papers and consultants that can
prove to you that their solution is the best; then have the competitor
dismantle their arguments.
Good luck! And by the way, I work in storage for a big company (I
won't tell you which one, nor which part of storage), and certainly
our storage system will have the lowest overall cost of ownership :-)
--
The address in the header is invalid for obvious reasons. Please
reconstruct the address from the information below (look for _).
Ralph Becker-Szendy _firstname_@lr_dot_los-gatos_dot_ca.us
| |
| Bill Todd 2005-10-24, 9:43 am |
| _firstname_@lr_dot_los-gatos_dot_ca.us wrote:
....
Typical rule of thumb is that operating costs
> (floorspace, power, cooling, excluding personnel, maintenance
> contracts) at least double purchase cost, often triple it.
> Administring a storage system (the personnel cost) is another huge
> factor on top of that. I would guess that administration cost could
> easily be 3x or 5x the purchase cost.
That certainly sounds like the right order of magnitude, anyway. I've
seen the claim that administration costs are 7x purchase costs in more
than one place, and if they lumped in operating costs with the former
the total would be almost exactly what you're suggesting.
Then again, administration costs *are* getting reduced by new management
technology and/or the development of systems sufficiently suave to
require relatively little human management (e.g., systems that can make
good use of a large group of disks which may vary in size over time with
minimal direction in the areas of performance and availability), so
these numbers may become obsolete before long.
- bill
| |
| Spindle 2005-10-24, 9:43 am |
|
dwhitlow1@wi.rr.com wrote:
> Hi All,
> What is the best practice for computing the cost per GB within a SAN?
> If only computing the cost per GB on a single IDE HDD, it's relatively
> simple. However, in a SAN environment, there is so much additional
> infrastructure including switches, management software, arrays filled
> with varying degrees of capacity, etc. Not to mention that fact that
> the array I buy today may not cost the same as the exact same array I
> buy tomorrow.
>
> Also, is there some standard for factoring the human capital involved
> as well?
>
> I'm doing a study internally for our company and looking for the best
> way to do this. The end result is to show the difference between
> varying levels of storage including FC vs. iSCSI-based storage, FC vs
> SATA HDDs, and things like that.
>
> Any tips on how others have calculated this would be greatly
> appreciated.
>
> Thanks,
> Don
You could find some inspiration in this Adaptec study:
http://www.adaptec.com/worldwide/pr...p_tco_iscsi_san
They were trying to prove that iSCSI is less expensive, so some bias
should be expected, but I haven't seen anything outrageous in their
approach.
Anyway, not taking performance into consideration makes for odd
comparisons, IMO, like putting side by side a racing car and a quiet
sedan.
I don't have a good suggestion on how to do that either, but try
looking at the SPC site for a hint
http://www.storageperformance.org/specs.
Good luck
PS: I wonder if you would consider sharing your conclusions when
done. Should be a great document.
| |
| Faeandar 2005-10-24, 9:43 am |
| On 13 Oct 2005 08:23:29 -0700, dwhitlow1@wi.rr.com wrote:
>Hi All,
>What is the best practice for computing the cost per GB within a SAN?
>If only computing the cost per GB on a single IDE HDD, it's relatively
>simple. However, in a SAN environment, there is so much additional
>infrastructure including switches, management software, arrays filled
>with varying degrees of capacity, etc. Not to mention that fact that
>the array I buy today may not cost the same as the exact same array I
>buy tomorrow.
>
>Also, is there some standard for factoring the human capital involved
>as well?
>
>I'm doing a study internally for our company and looking for the best
>way to do this. The end result is to show the difference between
>varying levels of storage including FC vs. iSCSI-based storage, FC vs
>SATA HDDs, and things like that.
>
>Any tips on how others have calculated this would be greatly
>appreciated.
>
>Thanks,
>Don
I have read articles from end users that put the
management/maintenance cost of 1GB of capacity at $55-$65 per year.
And when you consider that the 1GB one-time acquisition cost is about
$12-$14, well, that's alot of overhead.
I do not know if these numbers were for SAN, NAS, or other.
~F
| |
|
| Faeandar wrote:
> I have read articles from end users that put the
> management/maintenance cost of 1GB of capacity at $55-$65 per year.
> And when you consider that the 1GB one-time acquisition cost is about
> $12-$14, well, that's alot of overhead.
Both figures are highly variable, but the first more than the second.
The first is highly dependent on needed availability and protection -
i.e. you could just have hardware support @10% list price if you don't
care - which may well be the case for bulk storage of video, for
example, or situations where the information is so volatile that it is
better to re-run an experiment than to try make a backup between data
capture and analysis.
Thomas
| |
| Faeandar 2005-10-24, 9:43 am |
| On Fri, 14 Oct 2005 08:16:14 +0200, Zak <jute@zak.invalid> wrote:
>Faeandar wrote:
>
>
>Both figures are highly variable, but the first more than the second.
>
>The first is highly dependent on needed availability and protection -
>i.e. you could just have hardware support @10% list price if you don't
>care - which may well be the case for bulk storage of video, for
>example, or situations where the information is so volatile that it is
>better to re-run an experiment than to try make a backup between data
>capture and analysis.
>
>
>Thomas
Actually, those costs were based on things like backups (tape,
library, drives, etc.), cost of personnel to monitor said backups and
events, etc. I don't know if it took into account networking, power,
cooling, space, etc. though. It could just be the more visible
aspects.
~F
|
|
|
|
|