| Author |
Cost of storage calculator?
|
|
| lacka_dacka@yahoo.com 2006-04-09, 7:01 pm |
| Hi,
I am trying to come up with a calculator that lets me compute the cost
of storage per terabyte of data stored.
I have a 3 tier system (with various vendors) with a primary ->
secondary -> tape backup architecture
My goal is to see which vendor combinations are providing me the least
cost of storage.
I would like to know what factors to consider in creating my
calculator.
For example, even a simple list of items to consider will help. An
example list would be:
1. Disks
2. Ports on the switches
3. Licenses for each TB managed by the filer
and so on.
Also, it would help if someone can give me typical numbers to expect
for each item... for example, how much would one normally pay someone
like EMC for each TB stored... how do port costs work etc...
I really appreciate any help
/l
| |
| Faeandar 2006-04-09, 7:01 pm |
| On 5 Apr 2006 23:03:25 -0700, lacka_dacka@yahoo.com wrote:
>Hi,
>I am trying to come up with a calculator that lets me compute the cost
>of storage per terabyte of data stored.
>
>I have a 3 tier system (with various vendors) with a primary ->
>secondary -> tape backup architecture
>
>My goal is to see which vendor combinations are providing me the least
>cost of storage.
>
>I would like to know what factors to consider in creating my
>calculator.
>
>For example, even a simple list of items to consider will help. An
>example list would be:
>1. Disks
>2. Ports on the switches
>3. Licenses for each TB managed by the filer
>
>and so on.
>
>Also, it would help if someone can give me typical numbers to expect
>for each item... for example, how much would one normally pay someone
>like EMC for each TB stored... how do port costs work etc...
>
>I really appreciate any help
>
>/l
Non-acquisition costs are company specific. How much do you pay for
power? How much do you pay admins to manage this storage? Backup
admins to back it up? How much do you pay for tape and drives and
libraries? Do you have a factor for data center space per rack u?
Per cubic foot? etc.
As you can see there are numerous factors involved, it just depends
how deep you want to go.
Acquisition costs also vary greatly by customer. Company X may be
purchasing in round lots of 10TB a shot, while company Y may be
purchasing at 1TB a shot. Discounts will differ. This applies to
switch ports, management software, etc.
There is no industry calculator, but once you have all of your
specific variables you have a company specific calculator.
For me the cost of storage comes to about US$60/GB/year. But my
inputs will likely be different from yours.
~F
| |
| Ed Prochak 2006-04-09, 7:01 pm |
|
lacka_dacka@yahoo.com wrote:
> Hi,
> I am trying to come up with a calculator that lets me compute the cost
> of storage per terabyte of data stored.
>
> I have a 3 tier system (with various vendors) with a primary ->
> secondary -> tape backup architecture
>
> My goal is to see which vendor combinations are providing me the least
> cost of storage.
>
> I would like to know what factors to consider in creating my
> calculator.
>
> For example, even a simple list of items to consider will help. An
> example list would be:
> 1. Disks
> 2. Ports on the switches
> 3. Licenses for each TB managed by the filer
>
> and so on.
>
> Also, it would help if someone can give me typical numbers to expect
> for each item... for example, how much would one normally pay someone
> like EMC for each TB stored... how do port costs work etc...
>
> I really appreciate any help
>
> /l
Have you talked to someone in your company's accounting department?
This seems perfect for a little spreadsheet modelling.
HTH,
ed
| |
| lacka_dacka@yahoo.com 2006-04-09, 7:01 pm |
| Hi Faeandar... thats a good question... we have about 1TB of data and
we do have a datacenter. We are projecting that the growth will be
rapid this year and will be going to 10TB by the end of the year
because of an acquisition.
I am not sure how much we pay for power and space, but I will most
likely have to relocate the new storage to a different part of teh
building to make room...
Taking realestate and power out of the equasion, can you help me with
some ways of sizing the price? My question may be more basic than you
think... I am naive and even new to what the pricing models on a NetApp
filer may look like... how would I go about buying 10TB of space from
them? Do they license per terabyte?
/l
| |
| lacka_dacka@yahoo.com 2006-04-09, 7:01 pm |
| Faeandar ... your reply helped a lot... let me ask some more questions
on your deployment though... when you say $80 per GB... is that just
the disk cost or fully loaded (cost including your disk array, licenses
etc)?
Also, if you had to add another TB in your storage, would it just total
cost $800?
Also, given storage proces keep reducing, how much do you think the
same GB will cost next year?
/l
| |
| lacka_dacka@yahoo.com 2006-04-09, 7:01 pm |
| Sorry, I meant to say... at $60/GB right now ,would you expect to pay
$60,000 per TB?
Also, what rate of decline do you expect on that... what do you expect
to pay per GB next year and per TB next year?
/l
| |
| Paul Rubin 2006-04-09, 7:01 pm |
| lacka_dacka@yahoo.com writes:
> Sorry, I meant to say... at $60/GB right now ,would you expect to pay
> $60,000 per TB?
I don't understand this. 500 GB SATA drives are about $300 each at
newegg.com, 1000x cheaper than $60/GB. Of course you want to use
enterprise drives that are maybe $3 per GB instead of $.60/GB, but
that's still a huge ratio even after you add all the RAID and NAS
stuff. What's going on?
| |
| lacka_dacka@yahoo.com 2006-04-09, 7:01 pm |
| I have the same confusion when I started this project... I am guessing
it is the license cost we have to pay EMC and NetApp etc? Maybe other
hardware expenses? You need sometihng to plug the drive into
afterall...
/l
| |
| Bill Todd 2006-04-09, 7:01 pm |
| lacka_dacka@yahoo.com wrote:
> I have the same confusion when I started this project... I am guessing
> it is the license cost we have to pay EMC and NetApp etc? Maybe other
> hardware expenses? You need sometihng to plug the drive into
> afterall...
The key to the answer is that the cost was expressed per GB *per year*.
One can probably buy a pretty decent storage system for well under
$10/GB these days, but the purchase cost typically represents only a
small fraction of the overall expenditure: management costs are usually
ballparked at something like 5x - 7x the purchase cost (likely including
license costs but possibly not service costs), and license and service
agreement costs can easily more than double the raw hardware cost over
the life of the product.
- bill
| |
| lacka_dacka@yahoo.com 2006-04-09, 7:01 pm |
| Awesome Bill... thanks for that answer.... I find this to be a common
confusion for newbiews like myself... we tend to automatically just
pencil in the cost of the harddrive and then find our forecast very
screwed up.
| |
|
| Bill Todd wrote:
> The key to the answer is that the cost was expressed per GB *per year*.
> One can probably buy a pretty decent storage system for well under
> $10/GB these days, but the purchase cost typically represents only a
> small fraction of the overall expenditure: management costs are usually
> ballparked at something like 5x - 7x the purchase cost (likely including
> license costs but possibly not service costs), and license and service
> agreement costs can easily more than double the raw hardware cost over
> the life of the product.
And internal support/management si probably related to the number of
users. A video storage bank will have lower maintenance costs PER GB
than a database server or office data storage. Think about value of
data, volatility of data, number of users, changing requirements,
special software, etc.
Thomas
| |
| lacka_dacka@yahoo.com 2006-04-09, 7:01 pm |
| Does NetApp and EMC charge per GB/year?
/l
| |
| Faeandar 2006-04-09, 7:01 pm |
| On Fri, 07 Apr 2006 16:25:46 -0400, Bill Todd <billtodd@metrocast.net>
wrote:
>lacka_dacka@yahoo.com wrote:
>
>The key to the answer is that the cost was expressed per GB *per year*.
> One can probably buy a pretty decent storage system for well under
>$10/GB these days, but the purchase cost typically represents only a
>small fraction of the overall expenditure: management costs are usually
>ballparked at something like 5x - 7x the purchase cost (likely including
>license costs but possibly not service costs), and license and service
>agreement costs can easily more than double the raw hardware cost over
>the life of the product.
>
>- bill
Thanks for clarifying that for the group Bill, that is exactly the
point.
To really get a feel for storage costs you have to factor in the cost
of maintaining it over it time; usually acquisition (purchase) costs
are cheap comparitively.
Quick, simple example:
you purchase 10TB of storageX for $3GB.
Do you back this up? tape costs, library costs, people costs, offsite
storage costs
Do you make this highly available? datacenter costs, license costs,
monitoring costs
Do you replicate the data? license costs, capacity costs, network
costs
You can go on for a while like this. Just buying 10TB of disk is
simple, maintaining it over time is less so.
~F
| |
| lacka_dacka@yahoo.com 2006-04-27, 1:14 pm |
| Thanks Bill and Faeandar ...
My next question is then, would a product aimed at reducing the amount
of storage consumed by a company look interesting?
If so, how would you go about calculating its ROI? How would you
justify it and compute how much it could save you money?
Or would you just be uninterested in reducing storage consumption
because of some other reasons (like its not a big pain point)...
/l
| |
| Bill Todd 2006-04-27, 1:14 pm |
| lacka_dacka@yahoo.com wrote:
> Thanks Bill and Faeandar ...
>
> My next question is then, would a product aimed at reducing the amount
> of storage consumed by a company look interesting?
Perhaps not: most of the interest these days appears to revolve around
reducing the cost of managing whatever storage you need, not the cost of
the storage itself (which has been declining and is likely to decline
even more rapidly in the immediate future).
- bill
| |
| lacka_dacka@yahoo.com 2006-04-27, 1:14 pm |
| Thanks Bill... wouldnt reducing the amount of storage have an impact on
the cost of managing that storage? I mean... simply having less data
to deal with should have a significant impact on the bottom line
spending right?
/l
| |
| lacka_dacka@yahoo.com 2006-04-27, 1:14 pm |
| To add to my question Bill... I guess what I am getting at is... while
cost of storage itself is decreasing, isnt it the case that the amount
of storage an enterprise consumes is generally on a rapid increase?
And that offsets the decrease in the cost of storage...
So the product (storewiz) we are evaluating is one that would optimize
drastically (up to 60%) of the data entering the primary storage
servers...
The value proposition is that having less data implies less overall
costs wrt hardware and management.... and we for example dont have a
steady amount of data... its always (every quarter) growing by many
gigs (if not TBs)
/l
| |
| Faeandar 2006-04-27, 7:14 pm |
| On 27 Apr 2006 10:09:21 -0700, lacka_dacka@yahoo.com wrote:
>To add to my question Bill... I guess what I am getting at is... while
>cost of storage itself is decreasing, isnt it the case that the amount
>of storage an enterprise consumes is generally on a rapid increase?
>And that offsets the decrease in the cost of storage...
>
>So the product (storewiz) we are evaluating is one that would optimize
>drastically (up to 60%) of the data entering the primary storage
>servers...
>
>The value proposition is that having less data implies less overall
>costs wrt hardware and management.... and we for example dont have a
>steady amount of data... its always (every quarter) growing by many
>gigs (if not TBs)
>
>/l
In some companies that may be correct. In others the pure quantity of
data is not the issue, it's the management of that data.
For instance, having 100TB of data just sitting there is not really a
problem for me. But managing that same data ( association and
cleanup, SOX compliancy, etc) , is the hard part.
Most of "management" revolves around policies, which is not something
solvable by a product.
~F
| |
| Bill Todd 2006-04-28, 1:14 am |
| lacka_dacka@yahoo.com wrote:
> Thanks Bill... wouldnt reducing the amount of storage have an impact on
> the cost of managing that storage?
Maybe, maybe not. And if it did, there are certainly some ways of
reducing the amount of required storage that would *increase* the
absolute cost of managing it.
So it's arguably better to concentrate on reducing the larger cost
(managing storage) and let the (often far lower) hardware cost fall
where it may.
- bill
| |
| lacka_dacka@yahoo.com 2006-04-28, 1:14 am |
| So one example that we got from the vendor was something along the
following lines... sort of an ROI calculator (my original topic)...
In this case, the company was addng about 25 - 50TB a month of new
data... the company uses an FA960e (which costs about 225K with out any
shelves populated) and can host about 84TB max (raw space... logical
space when taking RAID1 into consideration is much lower... around
50TB)....
So for a company with growth like that, I guess such a product would
make sense? I guess... based on your comments Bill and Faeandar, would
it be accurate to say that reducing the amount of storage entering the
primary storage is only valuable if the company has a huge growth rate
of new data?
/l
| |
| Sto RageŠ 2006-04-28, 1:12 pm |
| Would happen to know how much this storewiz device costs?
What about the data that's already on the NAS before the storewiz
device is put in place? Looks like it only compresses new data flowing
through the device (on-the-fly compression/decompression), which means
one has to "empty" the NAS device first after intorducing storewiz into
the network and then bigin to "fill" it to get the benefit.
-G
| |
| Faeandar 2006-04-28, 1:12 pm |
| On 27 Apr 2006 23:51:17 -0700, lacka_dacka@yahoo.com wrote:
>So one example that we got from the vendor was something along the
>following lines... sort of an ROI calculator (my original topic)...
>
>In this case, the company was addng about 25 - 50TB a month of new
>data... the company uses an FA960e (which costs about 225K with out any
>shelves populated) and can host about 84TB max (raw space... logical
>space when taking RAID1 into consideration is much lower... around
>50TB)....
>
>So for a company with growth like that, I guess such a product would
>make sense? I guess... based on your comments Bill and Faeandar, would
>it be accurate to say that reducing the amount of storage entering the
>primary storage is only valuable if the company has a huge growth rate
>of new data?
>
>/l
My initial thought is that if you're adding that much data per month
you probably don't need high performance storage for all of it, so
your costs should significantly drop if you figure out what needs
performance and what doesn't.
Now if you really are adding that much high performing data then you
have a bigger problem.
I've talked to this vendor and they claim they only introduce minimal
latency but 1) I have yet to prove that, 2) the compression ratio they
use in their marketing is the absolute best case scenario, and 3) they
did not have failover capability when I spoke with them.
~F
| |
| Faeandar 2006-04-28, 1:12 pm |
| On 28 Apr 2006 08:23:55 -0700, "Sto RageŠ" <netbacker@gmail.com>
wrote:
>Would happen to know how much this storewiz device costs?
>What about the data that's already on the NAS before the storewiz
>device is put in place? Looks like it only compresses new data flowing
>through the device (on-the-fly compression/decompression), which means
>one has to "empty" the NAS device first after intorducing storewiz into
>the network and then bigin to "fill" it to get the benefit.
>-G
It can do a read/re-write of the data to compress in the background
with spare cycles. I beleive it is throttleable as well.
~F
| |
| lacka_dacka@yahoo.com 2006-04-28, 1:12 pm |
| Thats Correct Faeandar... it can compress existing data as well...
I agree that its implementation doesent fully match marketing yet...
but I guess my question is... *if* the product were to match marketing
and be able to introduce minimal latency and at the same time compress
atleast about 60%, would this be a type of technology to invest in?
I see all the arguments that I need to cut down on the cost of
management of the data... and beleive me, we are doing everything we
can to do that using products from Archivus etc... But in addition, we
do spend a lot of money on NetApp filers that these kind of products
seem to be able to help with... Not to mention, since I started this
thread, in doing more research into our environment, I am amazed to see
how much money we are paying on energy bills alone with all the storage
equipment we have...
/l
| |
| Faeandar 2006-04-28, 1:12 pm |
| On 28 Apr 2006 09:19:38 -0700, lacka_dacka@yahoo.com wrote:
>Thats Correct Faeandar... it can compress existing data as well...
>
>I agree that its implementation doesent fully match marketing yet...
>but I guess my question is... *if* the product were to match marketing
>and be able to introduce minimal latency and at the same time compress
>atleast about 60%, would this be a type of technology to invest in?
>
>I see all the arguments that I need to cut down on the cost of
>management of the data... and beleive me, we are doing everything we
>can to do that using products from Archivus etc... But in addition, we
>do spend a lot of money on NetApp filers that these kind of products
>seem to be able to help with... Not to mention, since I started this
>thread, in doing more research into our environment, I am amazed to see
>how much money we are paying on energy bills alone with all the storage
>equipment we have...
>
>/l
There is no one on this board that can tell you if it's worth it or
not. That's an individual requirements issue.
If all the things you mentioned are benefits to your company, why
would you not? That was not rhetorical, look to the reasons this is
not a good idea even though it meets all your obvious requirements.
ie.
do you need failover?
is it worth it to buy a storewiz per filer (since they only supported
one each when I looked)?
is that worth the management and potential issues that go along with
it?
Whatever other issues there may be...
It's something only you can decide, I think alot of pro's and con's
have been covered. You can always eval it too.
~F
| |
| lacka_dacka@yahoo.com 2006-04-28, 1:12 pm |
| I understand Faeandar ... I am just trying to see if I am investing in
the correct type of technology... I dont want to buy something that is
seen as non standard
/l
| |
| Faeandar 2006-05-01, 1:15 am |
| On 28 Apr 2006 11:41:01 -0700, lacka_dacka@yahoo.com wrote:
>I understand Faeandar ... I am just trying to see if I am investing in
>the correct type of technology... I dont want to buy something that is
>seen as non standard
>
>/l
Unless they are using some wild-XXX algorithm that no one's ever seen
outside of their company compression has been standard in many
environments since it was available.
And even if something is non-standard, if it serves a need then it's
worth it.
~F
| |
| goutham.rao@gmail.com 2006-05-01, 1:15 am |
| Awesome... Thanks Faeander!
| |
| steven.russo@yahoo.com 2006-05-04, 7:12 pm |
| Faeandar wrote:
> On 28 Apr 2006 11:41:01 -0700, lacka_dacka@yahoo.com wrote:
>
>
> Unless they are using some wild-XXX algorithm that no one's ever seen
> outside of their company compression has been standard in many
> environments since it was available.
>
> And even if something is non-standard, if it serves a need then it's
> worth it.
>
> ~F
This thread is timely. I was at interop this week and met with
StoreWiz. They provided me with their ROI calculator. We have about
200TB of data and grow at the rate of 20% a year. Plugging in my
numbers into their calculator, here is what I got
Start: 200TB of data
Growth Rate: 20%
Cost per terabyte to me: 12K just for hardware (fully loaded with
NetApp)
Assuming a 30% reduction in storage using StoreWiz compression
hardware, I get a net savings of 1.2M in 2 years. The cost of the
Storewiz appliance is 45K.
Given that I only computed in hardware costs per terabyte (we typically
spend more per terabyte than that if you factor in management,
electricty and floor costs), I think this is a very good deal
HTH
| |
| NoSpam 2006-05-04, 7:12 pm |
|
<steven.russo@yahoo.com> wrote in message
news:1146777669.786328.167770@i40g2000cwc.googlegroups.com...
> This thread is timely. I was at interop this week and met with
> StoreWiz. They provided me with their ROI calculator. We have about
> 200TB of data and grow at the rate of 20% a year. Plugging in my
> numbers into their calculator, here is what I got
>
> Start: 200TB of data
> Growth Rate: 20%
> Cost per terabyte to me: 12K just for hardware (fully loaded with
> NetApp)
>
> Assuming a 30% reduction in storage using StoreWiz compression
> hardware, I get a net savings of 1.2M in 2 years. The cost of the
> Storewiz appliance is 45K.
>
> Given that I only computed in hardware costs per terabyte (we typically
> spend more per terabyte than that if you factor in management,
> electricty and floor costs), I think this is a very good deal
>
> HTH
>
so 1 StoreWiz appliance can frontend multiple filers? or do you have 200TB
on one NetApp filer?
-G
| |
| Faeandar 2006-05-04, 7:12 pm |
| On Thu, 4 May 2006 16:25:00 -0700, "NoSpam" <NoSpam@NoSPam.com> wrote:
>
><steven.russo@yahoo.com> wrote in message
>news:1146777669.786328.167770@i40g2000cwc.googlegroups.com...
>so 1 StoreWiz appliance can frontend multiple filers? or do you have 200TB
>on one NetApp filer?
>-G
>
That's an excellent point, one I alluded to earlier. When I spoke
with them they could only front end a single filer at a time. And if
you want redundancy you have two appliances per filer head. At 200TB
of primary storage (non R class filers) you are looking at 3 filers
minimum, more likely 4+.
So if you have 4 filers, and want redundancy, you had to purchase 8
storewiz appliances.
And this also assumes their calculator is correct for your specific
data sets.
~F
| |
| steven.russo@yahoo.com 2006-05-05, 1:12 am |
| I currently have 200TB accross 6 filers.... and they seemed positive
that me that all our guys would need was just two storewiz
appliances... one of them being for redundancy.
The ROI calculator (which is a spreadsheet) seemed pretty basic. Email
me if you want a copy of the spreadsheet.
It basically shows how fast 200TB at 20% growth with 12K per TB would
add up every year with and with out storewiz compressing at 30%. They
also have a software tool that will accurately predict how well they
can compress (which I will give a try next week).
| |
| Faeandar 2006-05-05, 1:12 am |
| On 4 May 2006 17:50:23 -0700, steven.russo@yahoo.com wrote:
>I currently have 200TB accross 6 filers.... and they seemed positive
>that me that all our guys would need was just two storewiz
>appliances... one of them being for redundancy.
if this is the case then they've improved the product. When we spoke
with them about 5 months ago it was not the case. The storewiz
appliance was a one to one ratio.
they did say they were planning to support more than one filer behind
a single compression appliance so maybe they got that working. i
recall the appliance at the time had 6 gigE ports but could only front
one filer. i thought that was humorous.
~F
| |
| NoSpam 2006-05-05, 7:13 pm |
| <steven.russo@yahoo.com> wrote in message
news:1146790223.938090.293560@y43g2000cwc.googlegroups.com...
>I currently have 200TB accross 6 filers.... and they seemed positive
> that me that all our guys would need was just two storewiz
> appliances... one of them being for redundancy.
>
> The ROI calculator (which is a spreadsheet) seemed pretty basic. Email
> me if you want a copy of the spreadsheet.
>
> It basically shows how fast 200TB at 20% growth with 12K per TB would
> add up every year with and with out storewiz compressing at 30%. They
> also have a software tool that will accurately predict how well they
> can compress (which I will give a try next week).
>
Do you happen to know if it supports other protocols besides NFS and CIFS? I
am particulary interested in Snapvault and Snapmirror. The website says
iSCSI and FC coming soon.
I have sent them few emails and haven't received any response yet.
-G
| |
| steven.russo@yahoo.com 2006-05-15, 7:12 pm |
| They do not support Snapvault yet... It is only CIFS and NFS. What
deployment were you looking at this for?
It really only starts showing severe return on investment if you have
atleast 100TB of data and have a decent data growth problem (like
around 15 - 20%). Do you have that much data?
Steve.
| |
| NoSpam 2006-05-18, 1:12 pm |
|
<steven.russo@yahoo.com> wrote in message
news:1147732324.521938.117580@i39g2000cwa.googlegroups.com...
> They do not support Snapvault yet... It is only CIFS and NFS. What
> deployment were you looking at this for?
>
> It really only starts showing severe return on investment if you have
> atleast 100TB of data and have a decent data growth problem (like
> around 15 - 20%). Do you have that much data?
>
> Steve.
>
Here's where I see its potential use for us.
In our environment we use NetApp's SnapVault to backup our primary filers
to remote R200s. These R200s have a long retention period, like 14 nightly
and 13 weekly of snapshots. We also use their OSSV (Open System SnapVault)
to backup Windows and Unix boxes. These too have the same retention of 13
weeks.
These snapshots consume a lot of space, some times 300% of the source. This
is due to large log/text
files that get rotated on a daily basis on the source. These could easily be
compressed, so a device like Storewiz sitting before these R200 could
compress the snapvault deltas in real-time thus saving us a good chunk of
space on the R200s.Not sure how much these units cost though, but then if
they are like $100K each it would be pointless . We could as well add more
shelves and not deal with another piece of hardware to manage.
Yes, we will be reaching 100TB pretty soon on these R200s.
-G
|
|
|
|