Saturday, 24 January 2009

Sorting the wheat from the chaff: using ILM to increase Data Centre performance and make your data work for you

Because I can't find it on the web in its original form, here is the ILM article as I pulled it together in September 2007 (with much input from, in particular, Jan Zelezinski).


With growing pressures on the Data Centre to support organisations as they fight to remain competitive, and to deliver on an ever-growing list of security, compliance and environmental requirements, information lifecycle management (ILM) has never been more important.

At Logicalis, we think of ILM as ‘the right information to the right users at the right time and at the right cost’, which implies a comprehensive approach to managing an organisation's data throughout its useful life, from creation to deletion.

There's certainly plenty to discuss about the fact that, in reality, so little data ever actually gets deleted, but nonetheless ILM represents a fundamental approach to optimising performance in the Data Centre. However, it needs to be managed carefully to ensure that the benefit is clearly delivered to the Data Centre and organisation, not just in efficiency savings, but in appropriate application performance.

A piece of information being old doesn't mean that it is not vital for business planning, or required as evidence of compliance. An ILM approach involves a series of policies, procedures, practices and tools which align the business value of information with the most appropriate and cost effective infrastructure. This makes ILM a strategy for getting the best out of both the Data Centre and the applications it supports – not just a collection of technologies.

Unlike traditional hierarchical storage management (HSM), ILM looks at data from the business’ perspective – in other words, it’s interested in information rather than in data. It allows for more complex criteria for storage management than size, age, and frequency of access. But, once the organisation’s data is properly understood and classified, like traditional HSM an ILM approach will organise data into separate tiers according to specified policies, and provide for automated data migration from one tier to another. Typical storage tiers would range from fast, expensive and more energy consuming disk-based media such as fibre channel disk systems, through cheaper, but slower serial technology architecture (SATA) disks and tape-based devices, to offline tape in a secure offsite facility. The efficient selection of devices that results from such an approach can also make a virtualisation strategy more effective.

But there’s no magic wand - before any of this can happen, the organisation has to understand how its stored data translates to business value.

Delivering value from data

Stored data gives value in three ways: it provides the organisation with the operational information it needs to drive the current business process and to maintain productivity; it allows legal obligations to be satisfied; and the ‘knowledge assets’ held within it support decision making and innovation.

So the importance of any data does not rest solely on its age, or how often it's accessed. This is particularly true of compliance where the issue may be how quickly any one piece of information can be located at any one time. It's also true that faced with corporate governance regulations, it is becoming increasingly important to protect even seemingly worthless data from accidental destruction or loss. ILM users must specify different policies for data that declines in value at different rates. The structure, content and value of an organisation’s data will alter over time as business, compliance, application and technology needs change, so ILM policies must be adaptable if they are to remain valid.

Some of an organisation’s information is tidily structured in relational databases. However the majority of it will be in semi-structured, unstructured or non-electronic form, in content repositories, file systems, mailboxes, filing cabinets, and people’s heads. Information is useless if it can’t be retrieved; it is almost useless if it can only be retrieved by one person. The keys to the retrievability of semi-structured and unstructured information are the quality of indexing (metadata), the effective use of metadata by applications, and appropriate delivery of the information once identified. A traditional HSM strategy can help with the last of these, but it can’t do anything about the first two. ILM is a holistic approach to the needs of the business for information and knowledge, taking into account applications as well as storage technologies.

Getting ILM right from the beginning

Logicalis’ ILM roadmap assumes, first, the involvement of representatives of both the business and IT in the creation of an inventory of the organisation’s data (data types, metadata, content). What can be covered in a sentence as an idea can in reality be a huge project - the analysis and classification of large amounts of unstructured data can be daunting, if worthwhile.

We can then define tiered storage policies, and begin to implement them, using techniques such as storage path management to maintain appropriate information retrievability and application performance.

Once the organisation really understands its data and has begun to see the benefit of ILM in terms of Data Centre performance, we can start to grow the data’s business value and further tune storage performance with such mechanisms as enterprise search, data hubs, single instance storage, email archiving, and the introduction of selective archiving to ERP and other line of business applications.

The better we understand our data and metadata and the ways in which applications and users access them, the more added value we can gain, whether it be through improved data quality strategies or through newly discovered knowledge assets.

Undoubtedly, in a Data Centre where terabytes of data have to be backed up each day, where thousands of often concurrent users have to be supported, and where vital compliance information has to be identified and retrieved at the drop of a hat, implementing an ILM strategy can deliver a huge saving in hardware and maintenance costs - not to mention in power consumption. An ILM approach can put paid to the instinctive desire to store all application data on the same high-end disk arrays, and instead spread the appropriate data on to more cost-effective second, third and fourth tier devices. It will inform your strategy across the whole of the enterprise, and allow you to resist the temptation of purchasing the latest bit of kit that ‘would do the job’. The thinking horse should come before the technology cart.

This is a very worthwhile journey to take as part of creating a leading-edge Data Centre, with more and more UK businesses starting to embrace an ILM approach. The Information and Lifecycle Management Survey 2006, conducted by market analyst group Quocirca, found that, in the last twelve months, UK businesses have moved from the planning and implementation stages to fully incorporating ILM strategies as a core part of their infrastructure, with positive effects on the businesses they serve.

No comments:

Post a Comment