Background: Lundbeck manufactures drugs to treat conditions such as depression, schizophrenia, Alzheimer's and...
Parkinson's. The corporation has remote offices and subsidiaries in nearly 50 countries. At its Copenhagen headquarters, it runs an HP SAN with HP XP disk arrays for database storage plus EMC Clariion disk arrays for unstructured data. About half of its data is in the form of Microsoft Office files. In addition to that, there are lots of small applications used by the labs.
SearchStorageUK: Why has Lundbeck decided to implement ILM?
Per Ronnow Staffe: We have short-term aims and long-term ones. In the short term, we have a situation where data volumes are increasing at about 100% per year and if we don't do something, it will grow out of our control. We are not aiming to reduce storage capacity particularly, but to reduce backup windows and the cost of adding more backup servers. We want to reduce backup times from 48 hours at present for a full backup to 12 hours.
In the longer term, we want to categorise data so that it is stored appropriately to its use by the business. That part will be more difficult as it involves the line of business, which is not necessarily aware of what data classification is or how to do it.
SearchStorageUK: What is the lifecycle of data in Lundbeck's business? How do you expect this to change?
Staffe: We keep everything forever. At present, we don't really know how that will change with any proposed data categorisation. Above all, we want our data to be safe, and if we carry on as we are, we risk our data by our backups becoming unwieldy. At present we are backing up everything. What we are aiming for is to categorise data, so that what is not being used is not being backed up and that which needs to be retrieved in a given time can be.
SearchStorageUK: What stages will the project go through?
Staffe: The first stage was to describe the project. We started with the idea that it would be a traditional ILM project – data classification and the infrastructure to support it. But soon we realised we needed to add archiving to data classification and conceptually divided the project and formulated the aim of archiving all data and implementing a software solution to support that.
The next stage will be the selection of a vendor and we aim to do this by 1 July. We have approached five vendors with our project aims and we want to see what they will come up with. After 1 July negotiations will start with the chosen vendor with the aim being to close the contract by 1 August. The implementation will then start and we intend to be up and running by 15 November.
SearchStorageUK: What type of solution do you envision will result from the tender process?
Staffe: The fundamental requirements that we have specified are that any movement of data from primary storage to archiving and back from archiving to primary storage should be invisible to users and applications. From the specifications we have presented to vendors I expect them to suggest some kind of file area network (FAN) solution. Gartner called it a form of global file virtualisation in which you can spread files all over disk resources and present them all to the same view.
SearchStorageUK: How will you measure the success of the project?
Staffe: Key targets that we are aiming for are that the amount of data in our filestores should not be bigger than at 1 Jan this year and backups should not take more than 12 hours.
SearchStorageUK: What's next?
Staffe: Right now we're trying to get out of the habit of still working on old projects after starting new ones. So, we are running to finish everything by 30 June or we'll have to go to management to get more time – and that could be painful!