Storage Costs Average $25/GB/Month

The full cost of data storage averages $25/GB/month and is consuming 40% of IT operating budgets.

These figures appeared in an article on "single instancing" by Stuart Butts, CEO of data services company Xenos Group (@xenos_group). I asked Stuart to give Vertatique readers some insight into burdened data costs. His reply:

"The figure of $25/GB/mth was derived from a combination of customer engagements, analyst interviews and [published] reports..."

"...we talk about fully burdened cost, which is not only storage and backup but also floor space in the datacenter, staff to support, server infrastructure, electrical power and cooling. Also, for high availability operations a complete online duplicate infrastructure is required to support a 99.999% availability."

Media industry IT professionals have long understood the data storage cost and management problems inherent the combination of very large data sets and a workflow that require multiple format versions of each. Now, mainstream IT is looking at similar issues under the rubric of "single instancing".

As enterprises are becoming more aware of the true energy/carbon costs of 'cheap' storage, there is a growing interest in strategies that maintain only one copy of content that multiple users can share over time, with format conversion and redistribution occurring on demand.

One of the biggest excess storage culprits during enterprise content creation may well be email, where successive versions of an item are repeatedly emailed and returned during review cycles. This can easily consume storage across an enterprise 50X the size of the actual item, as well as obscure its canonic version.

Smart distributed file systems are ultimate answer. In the meantime, implementation of collaborative workgroup software and other systems that support pass-by-reference rather than pass-by-copy are now Green ICT tactics.

More about Green ICT and storage.

Update 2010.06.30
The European Research Consortium for Informatics and Mathematics' October 2009 issue of ERCIM News has articles about energy-efficient storage.

An Arguement Against Email Attachments

@ecologee wrote to me "Do you know that the SMTP protocol has an overhead of 30%? So it's really inefficient to share large files via email! You can simply test it: take a 10MB-file and mail it to a test-account, the mail size would be around 13 MB."

I found this explanation from Microsoft:

...SMTP uses an ASCII (text-based) format. Because most e-mail messages that are sent are text or Hypertext Markup Language (HTML) messages, very little content conversion needs to be performed on messages...It is important to note that SMTP does require about a 30 percent increase in size for binary content, such as executable attachments, because this content has to be encoded by using uuencode or BinHex. SMTP is not designed to transport large file attachments, because FTP is the preferred method to transport large file attachments. However, in many environments, a 30 percent size increase is acceptable.

This made sense until that last sentence. A 30% increase in file size means a 30% increase in energy consumption and CO2e every day/month/year that message is sitting on a spinning disk. Not to mention the purchase and deployment of more storage gear over time. Making a single copy of an asset available via a collaborative SaaS application or social media site instead of distributing many copies via email is both greener as well as more productive, since it reduces confusion over which copy is the canonic one.

IBM Suggests 5-Year Storage OpEX is 4X CapEx

Can u please send the analysis for the same

Hi Matt,
I am really interested in knowing how these numbers are alculated. If u could send me a detailed copy,I would really appreciate this favor.

Gagan

IBM Storage stats source

Hi, Gagan -

I got the slide from am IBM blog that cited some Vertatique work, but now I can't find the original IBM post. The slide is still being served up into Vertatique from this IBM site: https://www.ibm.com/developerworks/mydeveloperworks/blogs/optimbvaprogra...

Maybe you can track it down through there.

Good luck!

Matt

Carbon Footprint of a GB

@dcarli asks, "What's the carbon footprint per megabyte stored per month?"

GB is the new MB, so let's look at those.

I have a .5TB NAS device on the office network, rated at 5.6W. I ran the numbers and got ~11 grams of carbon per GB per month, with an electricity cost of 1/10 of a cent.

Of course, I don't know how representative my storage is of the global storage energy profile and a detailed analysis is more complicated.

Our device has a automatic sleep mode, rated at 0.6W. The drive is probably asleep or completely off 12 hours/day, which would cut our footprint almost in half, although this is an unlikely scenario for a 24/7 data center. this is On the upside, I'm not counting the carbon footprint of the backups, which in a data center might happen at least once a day. Or the data center AC. Or redundant storage, if required. Or the carbon footprint involved in the manufacture and ultimate disposal of the device, backup media, etc.

So, for the sake of discussion, let's say that a data center's number would be ~11g times a logical PUE of 1.8 equals ~20g/GB.

20 grams does not sound like much; its all those GBs in the aggregate that generates a lot of carbon. This is the materiality problem, which we saw clearly in the motivational analysis for digital delivery.

According to IDC, the amount of data created and replicated globally in 2008 was ~486 exabytes (~522 billion GB). At 20 grams per GB per month, that is a monthly storage footprint of ~5 million metric tons of carbon, or ~62 million tons/year. ICT is said to account for ~600 million metric tons, which would put storage at ~10%. These are all rough estimates, but give us an order-of-magnitude perspective.

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.

Post new comment

The content of this field is kept private and will not be shown publicly.