 |
||||
|
|
|
||
|
Contents 
Introduction
The long road to ILM
E-mail is the word
More than just messages
Learning to let go
Screensound builds an ILM
10 things to know about ILM |
|
||
|
|
||||
|
|
||||
Moving data from one tier of storage to another -- and, eventually, to tape for long-term archiving -- requires highly detailed, careful housekeeping. Conventional applications expect their data to be in a certain place, and if an ILM system is busily shuffling old data off to slower disks the applications will quickly become irate. Since the system won't work if the user has to hunt for the file, e-mail management systems must seamlessly change relevant links to reflect the content's new location. That requires ILM capabilities to be built deep within the enterprise application platform.
As ILM is taken to its logical conclusion, every document and piece of information produced in a business -- not just e-mail -- will need to be analysed, indexed, and moved according to policies that reflect its relative importance to the business (not just its age, as in HSM). It's a massively complex endeavour because the operating environment needs to know exactly where every piece of information is at any given time.
This means ILM will require that consistent data indexing and access techniques be integrated at every level of the enterprise architecture. In a fully implemented ILM environment, indexing will become a core function of data storage -- not something that's done once a day or week.
"It's not just about the spinning media that we're going to store things on," says Veritas strategic technical architect Simon Elisha. "It's really rewrapping the HSM discipline and marrying it together with the compliance discipline. Integration with e-mail systems has always been a core focus, and the next step is to take a level of integration and focus on data placement aspects, research, and recovery of data."
That Microsoft has backed away from plans to include its WinFS file system with Longhorn in 2006, confirms the complexity of this task -- WinFS's biggest selling point is its ability to continually index content, and to track that content between storage tiers.
Since implementing this capability at the operating system level remains complex, storage vendors will likely control the ILM agenda in the short term. So far, the most progressive vendor has been EMC, whose Centera content addressed storage array tracks every data element from the moment it hits the drives. Each piece of information is marked with a unique hash code based on the content of the information. This hash sticks with the data as it's moved between fast disk, slow disk, and tape.
Applications no longer worry about which specific disk or network volume contains a file; they simply ask for the file from the Centera cluster, which checks its hash index and pulls the information from wherever it's been tucked away. This is similar to how e-mail management tools virtualise the messages, indexing them in their own way and managing user requests for specific content.
The approach may be sound, but Centera is only one solution, and is tied to a specific vendor's products. In the longer term, the industry needs to both standardise methods for archiving, indexing, and policy creation so that enterprises can implement ILM solutions that just work. This requires standardised ways of describing and prioritising both content and the ways in which it is stored -- something that may in the long term be driven by peak body the Storage Networking Industry Association (SNIA).
"We're trying to work with vendors to explain where the drivers are, and explain to users what vendors are thinking about," says Ray Dunn, chairman of the SNIA's Storage Management Forum. "Information has different value for end users depending on how it represents the business. End users need to be able to think higher in the stack and think about their applications."
In other words, think not only about your storage infrastructure but also your content and how it's structured. Content management developers have joined storage vendors and infrastructure suppliers to contribute their own perspectives on the evolving field: knowledge and content management provider OpenText, for one, recently snapped up e-mail management vendor IXOS to enter the market.
6%
3%
