Six Steps to Building an ILM Foundation @ VIRTUALIZATION JOURNAL

Six Steps to Building an ILM Foundation

Inventory, classify, assign, provision, monitor, and chargeback

May. 3, 2005 10:00 AM

Digg This!

Six-Step Methodology for Building an ILM Foundation

This section details a six-step methodology for building an ILM foundation that establishes a clear picture of where your storage infrastructure is today, and gives you a clear process to follow as you move forward into the future.

1. Take Inventory of Existing Storage Infrastructure

Before beginning an ILM implementation, it's critical to understand what is in your current storage resource inventory. By figuring out first what types of resources you have and how much capacity is being utilized by different business units and business applications, you'll be able to utilize existing assets for your ILM initiative better, avoid unnecessary investments in new raw capacity, and ensure that your new storage tiers are being properly set-up to satisfy service level requirements.

The following questions can help you gain a thorough understanding of what resources you have and how they are being utilized:

How many and what kind of storage resources are there today, including DAS, NAS, and SAN?
What are the total capacity and utilization levels?
How are the resources logically and physically connected?
What are the application, host and user dependencies on each resource?
Which resources can be classified as Tier One, Tier Two, and Tier Three?
Which resources have excess capacity that can be leveraged for ILM?
What solutions are already in place for data protection and business continuity, such as types of RAID, mirroring, remote copies, and multi-path?

2. Classify Data Types and Map Data to Classification Model

"All data is not created equal." This is one of the fundamental premises of ILM, and requires that you understand the value of your data to determine where it should be stored, how it should be protected, and how much storage capacity will be needed for each ILM tier. While end users are best equipped to make value judgments about their data, surveying your entire end-user population to classify the value of every file is not practical. Fortunately, when it comes to classifying unstructured data, file-level attributes can be used to streamline this process. For example, every business application provides its own unique file extension - .jpg, .ora, .dat, .doc, .mpg, etc. - that can assist you in making judgments about data criticality and value. In addition, an analysis of when files were last modified and accessed can help you understand how important they really are.

Below are some questions you should answer to identify and classify your organization's data types based on its business value:

What kinds of data are in the enterprise?
How much of each type of data is there?
What applications and users created this data?
When was the data last accessed and last modified?
What is the criticality and value of the data?
What are the existing storage dependencies?

3. Assign Storage Resources to ILM Tiers

The third step in implementing ILM is understanding and defining service-level requirements for data access, recovery, and retention so that storage resources can be assigned to appropriate ILM tiers and data management disciplines can be incorporated that align business requirements with storage infrastructure.

Below are some guidelines for developing ILM service level agreements:

What SLAs are in place today?
Have SLA requirements changed?
What are the performance characteristics of your applications and storage systems?
What are the data retention and availability requirements of key applications?
How fast must different applications be recovered in the event of a disaster or unplanned outage?

Page 2 of 4 « previous page next page »

About John Kelly
John Kelly is Director of Product Marketing with AppIQ, and chair of the SNIA Storage Management Forum Requirements Committee. John has been active in the storage management market for over 13 years as an industry analyst and as Director of Product Marketing with HighGround Systems and Sun Microsystems. John has bachelor of science degrees in marketing and finance from Babson College and a master of business administration degree from Northeastern University.