Ok, so the use case for twitter as an effective tool for an extended dialog on the cost of enterprise disk, data management policies, and the upside of tape as a cheaper data retention medium was soundly disproved last night by me and my new good friends Phil Phil and the mysterious Techmute. Techmute apparently also has good taste in beer, but before I veer off on a complete tangent, lets focus in on data management policies and why they are important for your business.
Every non-IT business person would have you believe that any data they have every created has infinite value for all time. Because there is a perception that disk is cheap (it is, but storage management is not, at least not at an enterprise level) and every data center is a vast empty hall just waiting for more storage delivery (its not- common challenges of space, power and cooling abound in every company I've spoken to on the topic), we have created a society of data pack rats.
No one I know cleans up old data at any time for any reason. That marketing presentation from 2002? It might be good for a laugh, or to compare your powerpoint skills of today to where they are now, but that presentation has no business value. In fact, that presentation has either zero business value or even negative business value- it is occupying storage that could be used for something that is positively affecting your business, or perhaps it even has certain inaccuracies that could present a legal liability for your company (something about promises or capabilities not delivered, perhaps?). Data in your ERP system showing purchasing trends from 10 years ago belong in the Smithsonian, not as a valid business decision making information.
So what is the fundamental issue? The cost of disk decreases 20-25% per year. Most businesses' data is growing 40-60% per year. Your infrastructure team that supports storage is fighting a losing battle. There are technologies at hand to help- thin provisioning, data deduplication, archive and purge strategies, or even data subsetting. The solutions I am describing are a bit complex at this point, I admit, and possible topics for future blog entries.
Storage can be the single biggest line item for capital spend for companies.Take all those individual elements- the physical (storage, tape, optical, etc), and the logical elements (thin provisioning, data dedup, data subsetting, etc) and wrap them all together and what you start to construct is a framework for data management policy.
Keep in mind this is an opinion and not legal advice- your company's legal department should be able to guide you on what kind of data needs to be retained for how long. In essence, you need to work with your business partners and your company's legal department to identify what kinds of data you have, who owns that data (not IT) and what the company strategy is for data retention. Will you keep everything on disk so it is easily accessible? There is a significant cost to that. Will you introduce ILM (information lifecycle management) that determines how data moves from fast, expensive storage to cheap, slow storage to tape? Will you go to tape at all? Tape is one reason why the service is called backup, and not recovery. What is the likelihood that if you need that tape in 7 years for a legal review that you will even have a tape library or software that can read it? Will you include in your strategy the ability to move off old tape to new tape and keep compability/integrity with your now current backup solution? With that advent of cheap SATA disk and CDP (continous data protection) technologies, never going to tape at all is becoming a more and more viable option.
What is defining a data management policy worth to your business? If you could cut your storage and backup related infrastructure costs by 10%? 25%? 50%?, would it be worth investing the time in data management? The alternative is unrestrained growth and continuously chasing expansion of data center footprint and hoping that disk sizes continue to increase at a much greater rate than their weight, power consumption and cooling needs.
What are your companies data management policies or strategies? Can your company afford not to have a data management strategy?
No comments:
Post a Comment