The IDC expects data generated by enterprises to grow by 48% this year and for 90% of that to be unstructured data. At this rate, demand is far outpacing the reduction in the cost of storage. With an ever increasing squeeze being put on IT department budgets, how is this problem being tackled?
The issue draws interesting parallels to the energy crisis governments around the world are currently facing, with demands for more energy offset against rising fuel costs and environmental targets. There are some interesting parallels in the solutions being used by both as well and perhaps a few tips to be learnt.
Saying it with a smile
Power boards in the US have been using some well documented psychological principles to reduce consumption rates, with impressive results and best of all - they're incredibly simple to implement. It likely started with an experiment in 2004 conducted by Roberto Cialdini and the San Marcos power board on a few hundred randomly selected households in California.
The idea was to reduce power consumption by letting people know how much energy they were using compared to their neighbours. Phase one involved including information on their bills of average power usage for their area and ways to save energy. This caused people who were using more than the average to start using less, which was a good result. However it also caused people using less than the average to actually start using more - this is called the boomerang effect - people now felt justified consuming more. This was not so good. Overall it caused something called 'normalisation toward the mean' and basically meant that overall the energy usage for the area didn't change.
Phase two involved the same as for phase one, but it also included small emoticons - a smiley face if the household were below average use and a frown if they were above. This had the effect of lowering those above the average but keeping those below from increasing their usage. Simply by rewarding good behaviour to enforce what social psychologist call the 'injunctive norm'.
There were a few complaints about the frowns so when this was eventually rolled out they were dropped, keeping just a smiley face for 'good' usage and simply the facts for bad. This still maintained the desired effect, but with less whinging. The concept is now part of a number of 'green energy' initiatives.
OK, interesting, but what about our storage?
When we talk about unstructured data, 90% of the problem, typically we're talking about file shares and probably mostly user network drives or department folders. The places where data goes and very often never leaves. Some data are important, even if rarely accessed and makes an excellent target for archiving technology. However a lot of data either belongs somewhere else, are already copied somewhere else or are quite literally a waste of space and could just be deleted. The question is - how can you get users to tend to their often not so little patch of data dumping ground repository?
Make them pay!
One approach is 'chargeback'. This involves attaching a cost (e.g. per GB) of data and charging the individual departments directly for their use of space.
This however can be complex and time consuming. First you need to know exactly how much space each user is taking up. There are many problems, such as large numbers of files 'created by' the 'admin' of the filer which can make this a lot harder than it might at first seem. There are products available to help with this such as Symantec's Data Insight, which can give accurate owner information using inferred ownership techniques based on access behaviour tracking.
There's a larger problem though and that's taking this information, creating policies and then enforcing them. However a number of organisations have found it doesn't have to be that complicated. Sometimes just giving a simple indication of usage, commonly called 'showback' has been enough to make users take responsibility for their data and think more carefully about their usage. In some cases leading to the sorts of returns you'd expect of an all-out chargeback initiative. This way you can use the information from products such as Data Insight or Veritas Operations Manager Advanced without the 'policy overhead' of having to enforce anything and still drive desirable behaviour and cost savings.
Don't forget the carrot
So if you've been thinking about implementing a chargeback scheme but it's still on a 'someday' todo list because of the thought of having to produce and enforce all those procedures, maybe it's worth re-visiting the idea and seeing if maybe a soft push in the right direction might be enough… just don't forget what the energy companies learned and remember the carrot for those already making an effort.
If you are interested in generating usage reports or getting more accurate details on data usage, more information can be found on Data Insight for unstructured file share data here and on Veritas Operations Manager Advanced for structured/block data here.