So much is written about cloud this, cloud that, makes me wish for a sunny day. Its all so theoretical – did they tell you that? Yes lots of discussion about cloud storage, not so much practical application – at least on an enterprise scale. Yes there are the many home users paying outrageous rates to back up those precious family pics but most enterprises (I mean large) are still contemplating their navels about the application of cloud storage.
While enterprises meditate on cloud storage, data piles up, inevitably, undeniably, relentlessly. So although I recommend keeping that “cloudy” thought, instead, living in the here-and-now, consider the role of ILM in your storage environment. Remember ILM? Information lifecycle management. The storage industry was all a buzz about ILM a few years back. Not so much now. Now its more about storage efficiency, SSDs and cloud this and that. All interesting and useful, but while we wonder at these new technologies, data keeps piling up and screams for better management.
ILM is about knowing the value of data (aka, information – I won’t be splitting hairs on that point)and appreciating that its value changes over time. Of the many possibilities, value can be measured in required delivery performance, protection from loss, or sensitivity to prying eyes. Ultimately, knowing the value of data allows its placement on storage equipment best suited for its care and delivery.
Here’s what I say: its time to go back to basics and rediscover ILM: focus on the value of a dataset and what characteristics should be associated with it. And most importantly, the placement of the data onto the appropriate storage hardware. After all, remember we are in a recession. Optimum use of investment dollars on storage matters more than ever.
So while vendors push storage tier this and that, data deduplication, and all manner of storage technology, without a rational way to manage data, the vendors profit and your budgets get drained.
Simply identifying what data is in a data center is a good first step on the ILM journey. The identification process is not something that IT administrators can do behind closed doors. Business units must step up and own up to their data. After all, its the business unit’s data; who knows better then them what a data loss or poor delivery performance would impact. Is the data properly protected? and did the business unit let the IT folks know that it should be? Or is the business unit just hoping IT knows what to do?
In the absence of any input, an IT organization can treat all data the same from a delivery and protection policy point of view. Put all the data on performance storage and call it a day. This can certainly simplify things but its gonna cost. Its so much better to place data on the equipment that best meets it needs. Put performance data on performance equipment and archive data on, well, archive equipment.
But how in the world can a IT organization know what data belongs where? They can guess and use their powers of deduction, but I fear the folks eye’s in the risk management department would roll back in the heads if they knew that.
Here’s some ideas to get the ball rolling:
- Form a cross functional team to manage data across the enterprise. Make it permanent. Setup an operating procedure to inventory, audit and set standards for data as it moves through its useful life. There are some tools that can help but their capabilities are uneven to be kind – so I won’t mention any.
- Identify what data the organization, government entity or business depends on. Don’t expect IT to understand the value of data –that is a team effort.It might be surprising to find that IT is managing all kinds of extraneous data not knowing how important it is. Yes, IT can bring some tools to the effort that can help, but ultimately it will be the organization-government entity-business unit that must figure out what’s important.
- Characterize the data. Identify who is the owner of the data – or more accurately the information. Name, size, format, creator, retention policy, access restrictions, etc, etc, all need to be identified. Data characterization will become a very personalized activity. Each enterprise will have different sensitivities. The point is: associate metadata with the data. (sounds redundant). The metadata may be maintained by a management system or other means, but that’s not important - understanding the nature of the data is.
At least by taking these few meager steps, there is some hope to place data on the appropriate storage equipment. Although the appropriate storage technologies is a subject matter all in itself, at least a coherent scheme is possible to map data to storage once its characteristics are defined. That’s where storage tiers, HSM and all those other terms come to the fro.
I am not saying this is easy. It is not. It is, however, a continuous and necessary process not unlike managing the vital assets for an enterprise.
Turning our attention back to our navels, having walked through the data identification and characterization process, enterprises can consider placing data into a cloud storage environment based on its value. A process grounded in the knowledge of what data should go were.
Posted by Gene Ruth