ILM = Has It Losts its Meaning

View Comments

• ILM rebirth could come from a surge in creating metadata for cloud service. And with that a rejuvenation of Object Storage could be a huge contribution to computing. Secure and isolate-able storage paradigms in the cloud requires the function that OSD provides.

• Fun reading - thanks.

  ILM lives - it's just as relevent now as it was in the 80's (maybe 70's) when IBM delivered DFHSM to the mainframe world... given the explosion of data storage farms (I can recall when 2TB was a really large data center rather than next year's HDD), it may be more so now than then, though $70/MB was kind of a high price. :)

  ILM is a philosophy (there's a promotion for you). You can't buy ILM off the shelf, you have to embrace the concepts then decide, based on your own environment, which pieces go together to build the nirvana envisioned.

  BTW, I think your blog entry should be included in a BLM process - Blog Lifecycle Management - that way we could automatically schedule it to reappear every 12-24 months as a useful reminder that no matter how cheap (relatively) storage gets, we shouldn't waste it, nor the effort to manage it. Plus, we shouldn't lose sight of the fact that there are more economical ways to store information...

• Wow, OSD, when and where will we actually see it outside of slideware, visio, webex and at conferences? ;)

  Lynne you bring up a good point however about the need for meta data to add structure around all of the unstructed content. Ironicly, the value prop of unstrctured data is ease of use, deployment and associated things of not having to put data into a structured format. However with that goes the cost or complexity of having to after the fact wrap structure around unstructed data for management and other purposes.

  Rob, I like the BLM, great one :)

  Concur with the notion that buzzword burnout aside, ILM is very much a paradigm that goes way beyond DFHSM/HSM and thus not a product per say.

  I also subscribe to the notion that unlike many ILMers whose mantra is reduce the amoint of data being managed, something that is usually tied to a sales pitch of a product, to a different view. That different view is, store thus retain more data for longer periods of time, however, with a major caveat. That caveat is that more data is retained at a lower unit cost (people, hw, sw, power, cooling, floorspace, proteciotn, etc) without negatively impacting service levels. If that means moving data off to tape, optical, disk, removable disk, cloud, reducing data footprint via compress, dedupe, archiving among others, than so be it. Those are the techniques and technolgies that enable the expedition.

  Now on the other hand, if one can not accomplish retaining more data at a lower unit cost for greater data density, there is always the delete key!

  However, in between the delete key and out of control data growth, is some combinaiton of people, process, technology, techniques, metrics, tools and so forth for enabling denser, more effective and productive data infrastructures.

  Thanks for the comments and discssion.

  Cheers
  gs

• One of the great problems of ILM was the use of the term Archiving. It meant different things to different people, and many got confused. Then many early products failed to deliver on (sold/marketed) expectations. It is also a highly complex process, that requires input from the business.... something that often will not happen. In today's world, we all want set and forget ... with little initial input. After all, why can't the computer work it out for itself?

  Lastly.... it really doesn't fix many of the problems, especially at a file level. Using HSM (Storage Tiering / what ever) it may empty the disk of data, but you still have the file headers and, often, it takes more time to walk the file system than it does to move data... especially with backups.

  Object based storage with a global names space seems to me to be the way forward... but that'll be so long in the future!

• Greg,

  This is a great post and timely I believe as well. ILM has not gone away except from the current marketing vocabulary of storage vendors. For the customer, managing the information lifecycle is a very real need they continue to grapple with. The issue with ILM and the buzz that surrounded it in (2002/3)is the fact that it came from the storage industry. An industry that sells products, which ILM clearly is not. It is obviously a process. Anything with the term "lifecycle" in it is a process. There is no single storage product or technology that can satisfy the breadth of managing an information lifecycle. ILM in storage came to mean "tiering", which is simply data movement based on some value-based policy. There's a lot more that goes into an information lifecycle than moving data.

  I would argue that ILM is just as relevant today as it ever was, if not more so. The massive volume of new data being created by organizations requires a more holistic approach to information management that recognizes its lifecycle. That is, taking a top-down approach in understanding the type of information the organization creates, it's importance and how it needs to be managed and retained. Then determine the technology tools (applications & storage) best suited to meet those requirements. Otherwise it becomes "keep everything forever", which is simply not feasible. Today, more companies recognize this and I see more efforts to do just this in the ECM space; projects that are being called Enterprise Information Management (I know, great another TLA).

  I tend to disagree that the problem with ILM is its association with the term archiving. In its classic definition "archiving" is all about managing the information lifecycle. This phrase is well documented in the practice of archiving long before it was coopted by the storage industry. The problem is that the term archiving was ill-defined by storage vendors. Archive is not a product (technology) itself; it too is a process, i.e. backup does not equal archive, HSM does not equal archive.

  I also believe that object-based storage is the technology ideally suited for ILM. It was clear to me in the mid-90s when I ran several IT groups that the taditional file system was inadequate for long-term storage of unstructured data. IT only sees the volumes, folders, sub-folders and file names. It has no context to go with the data, which often only resides with the end user. Object-based storage allows content to be stored in context. Descriptive metadata enables intelligent, policy-based management, distribution and retention as well as context for search.

  Object-based storage is here today and in production around the world. This response wasn't intended to be a marketing pitch, but you can learn more about the technology at the Caringo website (try it free) and in a recent article in Storage Magazine (Dec 2009). I had wondered as well what happened to ILM, which I know is still an important organizational issue and it's good to see the conversation kicking back up.

  Derek Gascon
  VP Marketing
  Caringo, Inc.

• Thanks for the comments Siobhan and Derek.

  Cheers gs

• I think one has to separate the technology & the problem.

  The industry definitely has tendency to use too much hype around marketing terms (see : Cloud, Grid, ILM, HSM, CDP, etc...) without exactly knowing what it means, whether they actually need it & if at all how to implement it into an existing environment or a new one..

  Technology enables solutions that solve problem, that reduce cost.

  ILM is a basically a marketing name to describe an organization challenge to manage its data, while reducing costs for storing the data depending on its age and usage, and access policies.

  Now, these problems existing before they called it ILM, and they remain after the name has lost is market hype. The challenge still exists.

  The name popped, when the barriers of market solutions had to be broken, when organizations had to deal with more data at an exponential rate, just like cloud today is a market hype, of the fact that we need to use many servers to do a bigger computing/storing tasks, etc..

  There are various technologies, methods, workflows, to solve/improve these data mgmt / life cycle issues at organizations.

  HSM, for example, to my perception is technology (and again there are even many ways to implement such a technology), that enables to reduce costs by retained data in such a way.

  I think the name is irrelevant. What is interesting is the problems organizations are facing with managing their data today, and finding ways to help them do that, and reduce their costs. There are many solutions for many problem, not one for all.

  As the Zen phrase says : " The top of the mountain does not care from which direction you came to it".

  Ran