Have you seen the latest rhetoric from the storage industry? It comes with various labels such as "information life cycle management," "storage life cycle management" or "storage operations management," but it's really just the storage spin on utility computing. This storage automation vision states that corporate data requires different care and feeding through its life span and that storage hardware and software can support this through an automated, policy-driven infrastructure.
At this time, it's fair to view these vendor stories as nothing but hype. After all, these are the same companies that sold you on other fish stories and goofy acronyms. (Anyone remember EMC Corp.'s E-Infostructure or Cisco Systems' DEN?) Nevertheless, there's actually some reason and value here.
Today's storage infrastructure is really comprised of discrete silos. In a typical shop, the data center staff might own the back-end Hitachi Data Systems disks supporting Oracle, but network administration is responsible for the Network Appliance network-attached-storage (NAS) boxes, while operations controls the Veritas backup software and equipment. This situation creates a slew of operations problems.
None of the data in these systems is classified according to business criticality. Meaningless data is overprotected; other critical data is at risk. None of the equipment works together. Operating processes and procedures are repeated again and again. In the meantime, corporate data keeps growing and the IT staff falls further behind.
Your local storage salesperson will likely show up soon with a 20-slide PowerPoint presentation that parrots my last paragraphs and offers hardware, software and services solutions nirvana. While it would be easy to succumb to these offerings (accompanied by several rounds of golf no doubt), purchasing vaporware isn't your best option.
Let's get basic
Vendors are just beginning to discover storage automation. Some--EMC and Veritas Corp., for example--acquired many of the pieces and will need time to integrate companies and products. Others--Computer Associates International (CA), IBM Corp. and Hewlett-Packard Co.--are at the beginning stage of putting their solutions and strategies together. This early technology stage means there are no easy answers; the road to storage automation will be paved with hard work, lengthy projects and constant measurement. Progress begins with two preliminary exercises:
First, find every piece of storage equipment and understand where it lives in the company (both physically and organizationally), how it is used and who is responsible for its care and feeding. Dig deep here, as it's easy to miss important details. When was the equipment purchased? How far along is it in the amortization schedule? Which application is it associated with? Does the data need to be replicated elsewhere? What management information is available from the devices (e.g., utilization, performance, availability history)? These details are important because you want to truly understand the equipment, operating cost, staff makeup and enterprise interdependencies in order to cut costs and provide information-specific protection and performance.
Next, make sure you understand current and future business demands. Are there new business initiatives that will accelerate growth in the customer database? Will the company begin to keep storage-hungry reference data online? Are there any regulations that will mandate data archival? You will need to go through this type of analysis in terms of how it will impact each and every database and file system before beginning. This is necessary because you don't want to solve yesterday's problems--you want to anticipate tomorrow's needs.
Coordinate storage automation activities with the business unit and application folks before doing anything specific. Application managers are going through a similar automation and cost-cutting exercises with initiatives such as server consolidation, utility computing and organizational restructuring. Storage automation may fit into or overlap with these programs. To achieve the best results, overcommunicate with your IT peers and coordinate all storage activities within the larger IT context.
Attack pain points
At this point, you should know your storage resources, how they are used and what future demands to anticipate. The next step in the process is to outline existing and anticipated pain points. It's not good enough to simply state problems such as "too many incidents of storage downtime." Now is the time to really examine problems, looking for details that may cause or exacerbate each event. It's easy to say that a faulty Fibre Channel switch is causing storage service interruptions, but it takes a bit more digging to find out that the switch was misconfigured, and that the last upgrade didn't follow corporate change control processes.
Once you have a list of problem areas, separate them into three buckets: people, processes and technology. Rank these issues from most to least important. Don't forget to match each problem with its desired result by including the business, operational and financial impact.
For example, you may address a file system that continuously exceeds its storage threshold and crashes by employing a combination of streamlined provisioning processes, moving toward more automated tools and leveraging a pool of storage, classifying data and creating a data aging and archival policy. The impact of this fix could be enhanced availability of critical data (business impact), less administrative overhead (operational impact) and greater utilization of existing assets (financial impact). Quantify your goals and results when possible.
Once you have your list completed, it's time get busy automating storage tasks. Choose your starting point wisely--it can mean the difference between profitable success and miserable failure. You certainly want to address one of your more critical issues, but don't aim too high. Begin by tackling something small.
Storage operations process problems are good candidates because they won't require expensive technologies or time-consuming implementations. Use these simple issues to formalize the project management of automating: staff assignment, tool selection and metric creation. Make sure that you can achieve measurable results in a short timeframe, then communicate these achievements within IT and out to the business managers.
Finally, be honest with yourself. What went wrong? Where could the process be improved? Did you achieve your metrics? The goal should be to learn from mistakes and improve the process on an ongoing basis. In this way, the small automation problems you take on first will enhance your efficiency, as projects grow more complex.
Go through the list of process problems first, and then go to the people problems. As you move on to technology issues, invite some of your favorite established vendors and innovative startups to present solutions. Prepare these vendors by briefing them on your specific problems and the types of solutions you want to see. Include your requirements for scale, interoperability and device support. Winnow down your list by eliminating those that can't address your needs within six to nine months.
Again, start small and grow by using technology to automate in areas that offer the highest ROI, but don't forget to design an integrated solution for the long term. This will help support the progress you've already made while offering a technology platform that fits into your overall storage automation strategy.
None of this is earthshattering. No complex solutions or monumental ROI numbers, just solid IT discipline and project management. Storage automation is an excellent idea, but view it as evolutionary, not revolutionary. It's best to take matters into your own hands, identify things that need fixing and proceed with a prudent plan that addresses people, processes and technology. In this way, your company will see immediate and continuous improvements. Think of your vendor's future whiz-bang automation solutions as gravy on top of an already splendid meal.