ILM is Dead

by Administrator on August 29, 2006

I am flabbergasted. Befuddled. Confused. “ILM is dead” says a blog published on Computerworld a couple of days ago. Imagine such an article (a blog really) appearing in a publication that has run so many articles, advertisements, and advertorial projects extolling the virtues of ILM!

Hey Computerworld, why would you print such an article. Has the advertising dollar well around “ILM” suddenly run dry? :-)

We have been criticizing the marketecture around ILM since this blog first opened its doors. (We coined the word “marketecture” — by the way — to describe EMC’s ILM pitch, though I didn’t see any credit given to us in Robin’s blog!) It did not require a handful of surveys from SNUG to wake us up to the fact that the “Hype Curve” on ILM would quickly move consumers into the “Trough of Disillusionment” — which is what the post says has happened.

I would be careful not to throw the baby out with the bathwater. “Information Feng Shui Management” (the industry BS around the topic) is not ILM. ILM is a good idea, something that needs to be realized. ILM SNIA-style or EMC-style is just good old fashioned nonsense designed to sell overly complex and proprietary wares. We’ve said it before and we’ll say it again: ILM is more a people/process issue than a technology issue because it begins with the user — how he creates and names files so their worth is discernable and classifiable. It also requires a storage classification scheme so you can map platform cost and performance characteristics to what data requires in terms of access, update frequency/volatility, security, etc.

The blog incorrectly asserts that ILM is just HSM. Certainly, lots of products that call themselves ILM are just HSM (if even that). To do real ILM, other stuff, besides data age (creation date, date last modified, date last accessed) needs to drive migration through infrastructure. More granular information about data and its business context is required. HSM is a blunt instrument. Trying to do data management with HSM is like using a hammer to perform open heart surgery: messy and imprecise.

The right tools for ILM haven’t appeared in the market. And tools alone won’t create a disciplined managed data environment, in any case.

ILM is dead. Long live ILM.

{ 1 comment… read it below or add one }

Robert Pearson August 30, 2006 at 12:48 am

Well said! Jon…

Let”s Start Here with one ILM Strategy…
If you see value in ILM then your IT shop has a need for it.
Shops, and people, who do not see value in ILM have no need for it.

For years I designed and Implemented what I called “Rolling down and up the Storage Hill”. A lot of people thought I was being humorous?
One day someone asked me if I had ever heard of ILM.
Turns out to be a real nice sounding name for “Rolling down and up the Storage Hill”. I never looked back.
Some people would replace “down and up” with the better sounding “up and down’. I don’t because more Information goes down the hill than ever comes up in classic IT. There are some exceptions in the Web area.

Vendors are commission sales agents so they try to sell into the ignorance, not stupidity, of the clients who trust them.
ILM is not for everyone and even some of the people who need it do not have the infrastructure to support it nor the money to acquire that infrastructure.

As a rough rule of thumb you need an IT infrastructure with the bandwidth capability of 1 TB per hour minimum. This is from every possible source to every possible destination. This is why it is necessary to know, or find out, the “Speed Limit of the Information universe” for your IT shop. You would also find an I/O map handy.

Fortunately, not all of your Information needs ILM. Like HR Information, unless you are an HR Outsourcing firm. Then it is your life-blood and you need ILM if your stored Information exceeds the Storage capacity of your laptop.

The 5 Phase IT Lifecycle Management (ITLM) cycle is:
) Assess )
) Strategy )
) Design )
) Implement )
) Manage )

The “)” is used to indicate a circular process. My normal symbols do not process on this blog.

The IoD (Information on Demand) concept is the foundation on which ILM, and other processes, is laid.
My definition of IoD is:
IoD is a concept to achieve a Strategic goal. IoD requires the Pervasive Information Fabric using agents and multi-agents. To achieve this architecture the ILM (Information Lifecycle Management) phases of Strategy, Design and Implement must coordinate. The ILM phases of Assess and Manage are monitoring activities of the Pervasive Information Fabric.

E2EIoD is “End-to-End Information on Demand”.
I actually spent years working in obscurity developing this only to find IBM had done a better job and called it SOA.

Some of the pieces you will need are:
E2EIoD Information Areas and Processes
Information areas are:
1) Online
2) Nearline
3) Offline
Basic IT / E2EIoD / ILM Processes are:
1) Configuration
2) Replication
3) Migration
4) Re-Configuration
5) Redeployment

That’s all there is to it!

Thoughts for the day?
If you had IPv6 fully implemented so that each disk drive had a unique IP address could you Implement ILM better?
Could you Manage it?

Previous post:

Next post: