Without a sound management strategy, you could be wasting 73p out of every pound you spend on storage, says Danny Bradbury.
Data storage is now very cheap. Even when using hard disc storage rather than offline archival media, the price per megabyte has dropped significantly in the past few years. Perhaps it is not surprising, then, that many companies have become lax in managing their storage.
Businesses seem to have adopted the same approach to storage as software developers have to memory. Instead of having to eke out every last byte as they had to in the early days of computing, some programmers simply raise the minimum Ram specification for their application, confident that users' PCs will have enough memory to cope. Similarly, according to Jason Rabbetts, director of storage management consultancy Source Consulting, many IT departments are being drastically inefficient in the amount of storage they use.
Source Consulting offers a total cost of ownership modelling service specifically for storage infrastructures, using storage resource management tools to map out a data landscape for its clients. "We have discovered that while everyone is saying 'I need more and more storage', they are only using 51% of what they have," says Rabbetts.
Moreover, half of the storage used contains redundant data, such as temporary files or data that has not been accessed for more than a year. "It turns out that for every pound spent on storage, customers are putting 73p in the bin - 27% of their capacity is being consumed by data that is invalid," says Rabbetts.
Companies that lack a storage management strategy to make their infrastructure more effective are probably also overlooking other critical storage issues, such as availability, performance and cost.
Reducing IT overheads is a top priority in many companies at the moment. Increasing performance can also lead indirectly to higher revenues or profits through enhanced customer service or data mining processes. Increasing data availability enables a company's applications to make better use of its resources.
Mark Ellery, head of business development at Hitachi Data Systems, says the first thing his company does when supplying a storage solution is to conduct a workshop with clients to identify what they want from their storage infrastructure. It is only after completing this stage that he will attempt to get the support of senior staff for the project.
The third stage in creating a storage management strategy, says Ellery, is to design your infrastructure. This will enable you to revisit your financial case once you have a better idea of what is needed. The infrastructure design process is where much of the storage management strategy will become clear, as you define ways to automate critical processes.
For those companies that want to go back to basics, data modelling is a good place to start when defining your strategy. Working out how much redundant data you have and how volatile it is will help you to understand how your business uses its data, and which pieces of your data structure are most important.
At this stage, as you develop both your physical and your logical data models, you should be able to correlate them with the efficiency of your physical resources - which servers consistently run out of disc space, for example, and what data they are storing. Ask how important the data is to your business processes.
One storage trend that has attracted a lot of attention is the storage area network (San), which can help to improve the availability of your storage. Many companies still subscribe to a direct-attached storage model, in which storage devices are dedicated to a particular server. Sans move the storage onto a central storage network, which hooks them together using high-speed links (sometimes Fast Ethernet, but it could also be Fibre Channel or Gigabit Ethernet). Theoretically, all of the physical storage devices are then available to all of the servers.
In reality, interoperability has been a major problem for San users, because different devices from different suppliers rarely work together out of the box. Although things have changed a lot in the past year or so, you still have to be very careful about what you put together, says Rabbetts.
Virtualisation is one solution to the problem (see page 50), but Ellery says he has yet to see much interest in the new technology from larger corporates.
For small businesses or departments within a company, a network-attached storage (Nas) system could be a simpler alternative to a San. Unlike a San, a Nas device does not have its own dedicated network. Instead, it sits on the existing corporate local area network and contains its own file- serving logic, making it independent from any one server. In addition, interoperability is not as much of an issue with Nas, and the system will be much easier to install and manage.
A San or a Nas device may help provide the foundation for a good storage management strategy, but it will not solve all your problems. Many would argue that a San creates as many problems as it solves through its sheer complexity.
Whatever you decide to use, you need to identify problem areas in your storage infrastructure and cope with them, preferably pre-emptively. Storage resource management (SRM) tools are one way to help with tasks such as capacity planning. There is no shortage of tools on the market. IBM bought SRM company TrelliSoft in August and is now offering its Storagealert product, for example.
Fujitsu Softek's first application in the SRM market (launched in March this year) is a space management application. Such applications will become increasingly important as the size of your storage infrastructure grows. Nick Tabellion, Fujitsu Softek's chief technical officer, explains, "Let's say you have 100 servers. How do you know which one is filling up on a regular basis? When do you need to add more space to it?"
The management of such resources is often delegated to an administrator and forgotten. Consequently, the administrator becomes aware of a problem only when a departmental manager calls them to say that their application's disc is full.
Such space management tools can often produce capacity planning reports, conducting trend analyses against baselines that help administrators to predict when a particular disc array will run out of space. However, Tabellion would like to go further, introducing policy management techniques into his SRM tools.
Tabellion talks about action sets based on SQL queries that can be used to trigger events. For example, if a file system becomes more than 80% full, then he would like to be able to view all of the files that have not been accessed in the past 12 months and archive them using a back-up product, deleting them afterwards.
"That is a SQL statement that a person could generate with an intuitive graphical user interface," he says, adding that he is building "what-if" analysis tools into the product to ensure that administrators do not end up inadvertently doing something wrong and having a disastrous effect on the network. The key is to integrate the SRM and back-up software.
Archiving is an important issue, particularly for companies that are required by regulators to keep copies of data for a set period. Hierarchical storage management software can analyse use patterns, and archive the data that is accessed the least. "That is very different from deciding how you prioritise what to back up," explains Derek Warry, marketing director at storage management consultancy InTechnology. "You have to look at business-critical systems. Which of them have the most impact on your total business?"
Understanding which sets of data are most important to your business is a key deliverable of the data modelling process, especially when you tie your entity relationship diagrams - the logical model of your data structure - into your business processes. Tools such as System Architect, from data and business modelling tools supplier Popkin, can help with tasks such as these.
Storage automation is also prevalent in the tape market. Graham Hunt, OEM marketing manager of storage device maker Quantum's DLTtape Group, says the trend over the past four years has been away from direct-attached storage tape drives towards automated, centralised units.
Tape enjoys the same advantages today as it did historically: it is removable, which gives you advantages in terms of data security, and it costs less than archiving to hard disc. "Even if you do not remove your tape from your site, it still has a huge advantage," says Hunt.
A storage management strategy is one of the most difficult things to justify to the board, because it does not offer immediate top-line deliverables. Nevertheless, savings on storage devices in future budgets, along with increased availability figures, should go some way towards securing board-level buy-in.
Ten tips for managing your storage
- Understand what you want from your storage infrastructure before you start
- Consider a data modelling exercise so that you can better understand your existing logical and physical structures
- Use storage resource management tools to identify trends in the amount and type of data being stored so you can plan capacity
- Consider a storage area network or network-attached storage, depending on the size of your business and the power of your network
- Prioritise data for archiving and back-up using a mixture of data modelling and interviews with departmental managers
- Consider online remote back-up services if you lack in-house resources
- Examine policy management if your data structure is very complicated - it may help you to automate certain tasks
- Revisit your financial plan after designing your infrastructure so that you can ensure that your budget is accurate
- Highlight cost reduction, higher availability and resilience as drivers to get board approval for your storage project
- Expect to spend a lot of time and effort migrating from your existing storage infrastructure - unless you have a virgin site, it will be a complex task.
IMG identifies San as winning solution
IMG is a UK-based sports management company that represents sports stars from around the world. Its existing IT infrastructure consists of about 60 servers, each with its own storage attached locally. The amount of data stored across the company fluctuates between 0.5Tbytes and 2Tbytes, depending on the time of year and the number of sporting events coming up, and a storage management strategy was needed.
"Administering 60 servers is not the most efficient way of doing things," explains IMG's service delivery manager Ian Carter. "It is difficult to add more capacity to the direct-attached storage." The company looked at the storage market to find out what the most flexible solution was, and decided that a storage area network (San) would provide the necessary improvement in performance and reliability.
"When Sans first came out, they could not interoperate with each other, but all the big suppliers are now pushing the fact that they can interoperate. There is a common software layer," says Carter. He is referring to the Storage Management Initiative, created by the Storage Network Infrastructure Association and based on Bluefin, a storage interoperability proposal submitted to the organisation by various suppliers.
IMG has also changed its back-up strategy, moving away from server-attached tape drives spread across eight sites, each serving a local network segment. Carter says it took two or three hours every day to change the tapes and make sure the back-up happened.
Storage consultancy InTechnology sold the company VBak, an online back-up solution from Compaq. Now, IMG has a leased line between its core office and the VBak datacentre. Carter carries out a regular interview with each business unit and asks them which folders need backing up on their servers, coding that information into a server-side script that is used to carry out the back-up process.