Most IT managers now recognise that the cost of managing your storage environment will far exceed any costs involved in the acquisition of the storage hardware. The general ratio of management costs to hardware costs varies widely depending upon the user's environment, the number of different server OSes and the number of internal and external users. Typically it falls within a range of 2:1 to 8:1, although the evidence supporting those figures appears to be mostly anecdotal as the rapid change in the storage environment hinders accurate assessment. It also depends on what is included in the calculation. Ideally such a calculation would include software costs, people costs, and all the non-hardware-related goods and services that keep the data flowing and available to those that need to access it.
What is frequently overlooked is the cost penalty that is paid if data that is requested is not available.
What this really means is there are three distinct cost elements that need to be included in the calculation: the capital cost, the operational cost and the opportunity cost. Typically, we focus on the capital or acquisition cost and may do some "back of an envelope" assessment of the operation cost, but ignore the opportunity cost.
If we can't be confident that we can measure it and are not sure which pocket the funds will come from to pay for it, how can we work out if we get good value for our money? How do we know whether we are getting what we pay for?
The answer, of course, may depend on what we think storage management software should do and what you buy. Essentially, storage management software does many things. These may be placed in three categories. We have to manage all of the storage devices. We also have to manage the data that is stored on these devices and we have to manage the users that seek to access and use the data. In addition, the storage management software has to work closely with the applications that are being used.
Consider what we expect of storage management software. Each of us has special requirements, but it's a safe bet we can all agree that what appears below should be included in any reasonable, high-level guideline for an ideal storage management environment.
* The ability to grow the system transparent to the users without impacting overall performance.
* Automatic discovery of any changes in our storage resources, with appropriate responses initiated as needed.
* Make the relevant data available to any application whatever the particular OS running on the server.
* Monitoring of both the transactions taking place and all the constituent parts of the storage environment.
* Management of all those things listed above, and of their interactions.
* Optimisation of the entire storage system so that loads are evenly balanced, data is intelligently distributed, backups are unobtrusive and recoveries nearly immediate, and intelligence flows unimpeded by suboptimal nodes or connections.
* Predictive capability that lets us anticipate hotspots or breakdowns.
* Automation capabilities that take over much of the day-to-day operations, and hopefully are heuristically based, so that they get smarter as they are used more.
* Control of the overall storage system whether it be in a single location or distributed across a considerable geographic area.
* Disaster planning, response and recovery for when a fire occurs, a building contractor cuts a cable, or the user accidentally deletes or destroys a file.
That, more or less, is what we want - all closely enough coupled so that we can expect interoperability between the modules, and all administered through a limited number of management consoles. What we have in place may be a different matter altogether. Unless the issue is regularly reviewed and appropriate investments made it is easy to have the worst of all possible worlds. All too frequently we find there is a hidden penalty to be paid by organisations that run various versions of NetWare, OpenVMS, Windows NT/2000 and perhaps two or more versions of Unix. We may have been coerced to buy multiple software packages to address each requirement, one for each server operating environment and then specially adapted for the multiplicity of storage devices.
It's unfortunate but true. One of the first things we learn in IT is that the hoped-for economies of scale we seek, as the amount of information we need to store and manage grows, may often be nonexistent. It is likely to be far worse if, in the course of scaling, we engage with multiple vendors or try to solve our problems through the use of multiple operating systems.
At best we will look for all or most of the required capability in one package. Alternatively, we may have to settle for a limited number of software solutions that cover all of the requirement.
Given this complexity and the apparent costs involved, whether these are obvious or hidden, it behoves us to have another look at the bigger picture, do a better assessment of our current costs and future requirements and then to develop an action plan to overcome the deficiencies.