The redoubtable John Webster, storage analyst extraordinaire at Illuminata, penned a short defense of Information Lifecycle Management last week. It was so humble that I just wanted to take this frail orphan named ILM home and feed it sparrow’s milk with an eyedropper.
Call me a sentimental fool. When a defense of ILM begins
Is ILM dead? I don’t think so. Will ILM die before it even gets a chance to stand up and walk? No. However, with the way some storage company marketers are behaving lately, one could get the impression that ILM doesn’t exist anymore.
I start tearing up.
God bless you tiny ILM
John concedes there is no definition of ILM and that vendors appear to be abandoning the concept, at least as a major theme. I can only hope StorageMojo’s unrelenting criticism (see ILM is Bunk and Gee, “Users Cite ILM Shortfalls” – Maybe ILM IS Bunk and ILM: rest in peace) of this fatally flawed excuse to sell overpriced storage gack had something to do with it.
ILM isn’t dead, it is only HSM on a respirator
I was talking to Dave Patterson, one of the inventors of RAID, last week for another project I’m working on, and we got on the topic of 2.5″ drives. He commented that laptop drives are perfect for archive purposes because they are designed to power on and off many times. Load your data, turn off the drive, and only spin it up when you need to read something. Disk archiving that is faster and smaller than tape. An ILM/HSM store that makes sense for our new FRCP rules.
Storage Marketing tip: try adding value
John closes with some sincere advice for “big-bucks marketing pros”:
What to do with ILM now that it’s lost its hype potential?
If you’re a storage marketing pro, you only have two options:
- Find a new buzzword. Repackage ILM in a new wrapper. As you do, bring the new buzzword closer to reality. Make it something one can do now, something real, something concrete. Away with the ILM that’s a mere vision of the future. This is risky, however. Storage administrators are doing something they call ILM. And the buck doesn’t stop there. CIOs and other corporate players (legal, security, and records and information management professionals) are becoming engaged in ILM projects along with corporate IT. As a marketer, you might have privately thought that ILM was just hype, but you would certainly never have wanted to send that message publicly. Change horses midstream and you risk sending exactly that message.
- Hang in there with ILM. Yes, it defies definition. Yes, it looks different with every customer implementation. And yes, it feels shop-worn. But it’s way better that than smoking up the atmosphere with mo’ better hype.
And if you’re a storage administrator, just keep on doing that ILM thing — whatever it is.
If that sounds suspiciously like another American marketing campaign (tag line: “Stay the course”), you have a point. ILM makes as much sense as HSM ever did, and if it hadn’t been freighted with cosmic strategic significance by EMC and others (didn’t Sun briefly have a VP of ILM?), you wouldn’t have these corporate players sitting down to watch this much-hyped IT project go thud.
Half of all storage marketing is below average
And some of it is in the bottom 2%. I agree with John: we don’t need more hype. How about using the massive resources of the major players to actually help customers understand and deal with the important metrics of their data stores? Starting with a recognition that data is getting cooler and big iron arrays are economic for an ever smaller fraction of all stored data. Oh, wait a minute, do something original? Too many variables!
Hype is easier.
Comments welcome, as always. I moderate comments to keep comment spam under control – the run rate is about 50 per day – and vigorous discussion is welcome.
SNW is next week. I’m sure you’ll get at least 2 new acronyms 😉
I am not sure why you have this reaction to ILM?
John Webster and I must have had very similar experiences because he and I are in very close agreement on ILM.
ILM and HSM are different processes and they serve different business needs.
You absolutely need some form of HSM in an IT shop. It is the easiest way to do bulk migrations.
You may need ILM if you have a “business” need for it.
When I first came up with my precursor for ILM called “Rolling Down the Storage Hill” it was to solve the problem of successfully deciding which Managed Units of Information to retain (migrate) and which to remove (migrate down the Storage hill, archive or delete) as the Managed Units of “Enabling” Technology reached end of life and were de-commissioned. To be replaced by newly configured Managed Units of “Enabling” Technology.
Longer life-cycles become ILM candidates because the “Defined Persistence” or “Expiration” in the SLA (Service Level Agreement) requires these Managed Units of Information be present on the new Managed Units of “Enabling” Technology.
It is not always obvious to IT what stays and what goes. You need a Strategy and a defined process to be successful with these migrations.
In the case of multiple local or remote, or both, sites you need defined
replication. First for the bulk initiation and then for rapid synchronization.
In the old days of a master site with one or more secondary masters nothing was ever in sync. Primarily because of slow telephone lines. Not a problem anymore.
You cannot implement ILM, or anything like it, until you do your homework and
can answer questions about the finger-on-the-pulse, the heart-beat, the
Infotone, Content, ROI/TCO, the Speed Limit of the Information Universe, and the Pervasive Information Fabric using agents and mujlti-agents to deliver IoD in your shop.
I believe that if you are not doing Records and Information Management (RIM)
your need for ILM is low. A kludge of HSM, Backup and Archive softwares with
“glueware” scripts will probably work, as it always has.
Think about CMS (Content Management Systems). Most people have a totally
different functional definition than I do. CMS to me is a means of identifying
automatically those Units of Information that generate 80% of your revenue and which, if you lose them, will put you out of business. These may not be exactly the same sets. They are vastly different in an eCommerce, eBusiness enabled company and a traditional insurance company.
Final note. Tags and tagging as Information is entered into your IT shop is a
way to define Content as it is created. A form of mashup could be used to
describe the meaningful relationships between the entities, objects or Units of Information. If the entity, object or Unit of Information is determined to be
important as either a revenue producer or essential to staying in business it
becomes a Managed entity, object or Unit of Information. By definition a Managed unit of Information must be “Enabled” by a Managed Unit of Technology. Managed means an SLA (Service Level Agreement) is created, assigned and acted upon.
Robert,
Thank you for the thoughtful post. I think a longer response than I can provide here is probably needed. But I’ll try anyway.
The secular trend is that absent a compelling need to get rid of data, data gets retained. As storage gets cheaper, more data gets retained. I believe that our current model of storage arrays is fundamentally broken, that virtualizing “disks” no longer makes sense, and that as a result we are paying much more for protection and performance than we need to. As a result of the expense of storage, there is pressure to “manage” something that, fundamentally, doesn’t justify item-by-item management unless automated at very low cost. Which, despite some great technology out there, doesn’t seem to be in the cards today.
While I understand there are important record management issues for business and legal reasons, the constraints imposed by our storage infrastructures should be secondary – i.e. business needs should come first. We don’t worry that laptop CPU utilization is only 3%, because the value of a portable information appliance is so obvious.
Justifying ILM or tiered storage because of cost is putting the cart before the horse. What vendors should be doing, and what the next decade’s EMC stock-rocket will do, is figuring out how to enable massive storage at low cost. Which is why I find alternate storage architectures so fascinating. It can be done. Some vendor just has to decide to undertake the multi-year campaign to make it happen.
I think your answer is a very fine answer. I will make a note of it to use it for future work.
I believe you are too modest. Your answer applies to many processes, features, functions, needs, wants and desires of IT. ILM is tiny part of that puzzle. I use ILM as one tool to give people the choice of Managing Information at one of the critical junctures in its life-cycle, migration.
I am a lot more concerned about Findability. In particular the SFO (Search, Find, Obtain) function, which is what the “Enabling” technology infrastructure must have to deliver Findability. Close on the heels of Findability is the User Experience (UX). Past these is Content Management. I can do some things with HSM, ILM, tiered Storage, VTL and a host of other nice ideas but none of them gives me that “Fundamental Shift in the Paradigm”. After all these incremental changes you wind up back at square one.
I am sure we are more agreed than not here.
Speaking prophetically, once SOA begins to happen, a lot of wheels will fall off. Any vendor who is not “SOA aware” will fall by the wayside. This may take longer to happen than I am projecting at the moment.