|
|
|
|
||
|
||
|
|
 |
Home >> White Papers >> Simplifying Storage Management in Ten Easy Steps Simplifying Storage Management in Ten Easy Steps Storage decisions have become increasingly complex. In fact, many storage solutions today are more complicated than the problem they're trying to solve. How can you simplify the solution without compromising on the promise of increased control, scalability and reliability? How can you design and deploy a storage solution in less time with fewer resources without compromising on performance and manageability? The answers are closer than you think. It’s time to cut through the hype, and take a fresh look at the problem using these 10 easy steps.
Step 1. Is Managing Your Data Becoming More Complex Than Managing Your Business? Storage Area Management (SAM) solutions are ideally suited to solve complex storage management problems of very large or distributed data centers with outdated topologies already in place. In these environments technologies such as storage virtualization, Fibre Channel connectivity, and management software are a necessary evil because despite their complexity, they are easier than managing what is currently in place. But because these solutions were designed to solve complex problems, the solutions themselves are also inherently complex. Consequently, these solutions are not the answers to everyone’s problems. For the vast majority of medium-sized companies, SAM solutions are far more complex than the problem. As a result, companies can invest considerable resources deploying and managing complex SAM software, switches, and elaborate storage architectures, without any real sense of whether tasks like monitoring, managing and backup are easier or more efficient than before. So are there ways to take the complexity out of deploying and managing storage systems that meet today’s demands, and still offer a reliable platform for growth and expansion? Quite simply, yes. Surprisingly, the answer to many of today’s medium and large company storage management issues lies in the hard disk array itself. Performance disk arrays take much of the complexity out of managing data because many of the performance, compatibility and management capabilities are built right into the array. It does this without using software that needs to be loaded onto servers, networks or switches. So before you look at adding complexity between the disk array and the user, look at what capabilities are inherent in today’s performance disk arrays. You may find money spent on the disk array rather than SAM software and infrastructure will give you the greatest return on investment. Step 2. Don’t Add More Technology Layers Than You Need Just a few years ago, before the emergence of Storage Area Networks (SAN) and Network Attached Storage (NAS), we routinely used storage systems with minimal layers between the servers and the disk array. And, somehow we managed to safely and securely store, backup and manage this data cost-effectively. Then things changed. Many would like us now to believe that we need new layers of complexity between the servers and the hard disk array in order to manage data properly. This includes layers of connectivity with switches, layers of manageability with Storage Area Management software, layers of scalability with Fibre Channel connections, and layers of storage optimization with virtualization. However, with each of these layers, we’ve added more work and more cost. Have some of these new technologies improved redundancy or data reliability? Perhaps. But only with considerable added complexity. Now we have three to five times as many drivers, protocols, switching layers, equipment and standards to maintain, requiring more and more highly trained specialists. For these reasons, data management and administration costs are estimated at over 70% of overall storage costs versus only 12% for the actual hardware.
If you want to simplify your storage systems, think of how to minimize the layers of complexity between your users and your data without sacrificing the control and visibility you need to manage company data. In turn, you keep system management less complicated and less volatile by avoiding technology layers that can potentially decrease performance, increase network traffic, introduce potential driver or operating system conflicts, and add new components to test, track, manage and upgrade. Step 3. Build Storage Systems To Be Flexible With Open Storage Portability. Today, storage systems can be configured as Direct-Attached Storage (DAS), SAN or NAS. Who knows what options will be available tomorrow. For this reason, the hard disk array – the heart of your storage system – should be entirely in your control and should be open systems compatible to work with any operating system, any platform, and any storage architecture. In other words, you should be able to migrate your disk array from a DAS to a SAN or NAS and vice versa whenever you need.
In the diagram above, you can see how the same disk array can be attached as a DAS, SAN or NAS – or in any combination of environments – at the same time. You should also be able to add or rearrange capacity at your convenience, do your own rebuilds and drive mappings, and monitor and manage the system without complex software. Although this may seem obvious, with many solutions this is not the case. Unless disk arrays are designed with simplicity in mind, they likely will require tweaking, testing or multiple vendor interaction to change a configuration. This may include new drivers for the server, operating system upgrades, new management software versions – even reformatting the data if it was stored using a proprietary format. If you’re looking for simplicity, expect the hard disk array to be pre-configured and ready to port between topologies without upgrades. Scalability. Disk arrays need to be able to meet your foreseeable storage needs. IDC has forecasted data growth to increase 90% through 2004. So you need to reasonably expect to double your capacity needs every 1 to 2 years. Although each company and environment is different, some applications are, in general, growing faster than others. The more of these applications you have, the more important it is that you deploy a system that allows you to expand at your pace, and on your schedule. According to Merrill Lynch, some of the most storage intense enterprise applications and externally networked applications include:
Open Systems Compatibility. Today’s best disk arrays are designed to be open systems compatible. This is a departure from proprietary designs that required special drivers or compatibility kits for each system. These proprietary designs add unnecessary overhead to operating systems and networks, and introduce additional points of failure into the system. Worse, proprietary disk arrays only work on one platform, and cannot be moved to another platform, i.e. Sun only works on Sun, HP only on HP, IBM only on IBM, Compaq only on Compaq, and Dell only on Dell. Open systems disk arrays are less complicated because they include the operating system and software compatibility they need right into disk array. While certainly important for initial deployment, this is even more crucial down the road when you want to connect this disk array to a new server or operating system, or change to a new topology. So regardless of whether the disk array starts off as a DAS, SAN or NAS, if you want to make your life easier today and tomorrow, be sure to start with an open systems compatible disk array. Step 4. Build From A Solid Base At the heart of any good storage system is a reliable, high-performance hard disk array. It doesn’t matter what RAID level you use, how many LUNs or partitions you create, whether you connect it via SCSI or Fibre, or whether its DAS, SAN or NAS. Just make sure that it is designed and built with the highest quality and highest performance components, including the drives, bus, controller, and enclosure. When it’s your assets on the line, get performance and reliability information up front from your vendor. If these performance and reliability metrics aren’t critically important to your vendor, it’s probably the wrong vendor. Appendix A offers some performance and reliability specifications you should use when considering a RAID disk array. Step 5. Understand When And Where Speed Counts As enterprises experience tremendous growth in their storage capacity needs, and deploy complex topologies and architectures to scale capacity to meet this growth, the impact on performance has become paramount. According to AllSAN Research, the amount of corporate data is doubling every six months. By the end of 2002, enterprises are expected to consume 1.4 million terabytes of storage annually – 12 times the amount consumed in 1998. In addition, Forrester Research reported that 30% of enterprises surveyed are consuming six or more terabytes of storage per year. Clearly, the challenge has become increasing capacity without decreasing performance. Many storage vendors would like you to believe that scalability and performance are synonymous. They are not. Storage solutions – including the disk array, connectivity, routing and management – need to be specifically designed this way. For instance, depending on your storage architecture, there can be a variety of connectivity and routing layers such as networking servers, data and application servers, virtualization servers, NAS filers, and switches that all have an impact on overall system performance. And naturally, the more of these layers you add, the slower the performance. This is one place where the right disk array really pays off. If the disk array has built-in capabilities to multi-host, support multiple vendors, and support multiple topologies simultaneously without drivers or compatibility kits, it minimizes the need for these layers of complexity and, in turn, improves overall system performance. In some storage solutions the disk array itself is the bottleneck. Even though today’s high-end RAID disk arrays have impressive components such as…
…this alone is not enough. The Predictive Performance Scalability diagram below illustrates how large monolithic arrays – which use one controller across multiple disk arrays – add only incremental performance for large additions of capacity. In contrast, modular disk arrays – each equipped with high performance RAID controllers – add incremental performance for each incremental addition of storage. In short, this illustrates how modular disk arrays allow capacity and performance to scale proportionally, for more predictable and consistent system operation.
Step 6. Invest In Investment Protection When you talk about investment protection, people often think of Return on Investment (ROI). ROI is important, but investing in storage systems is more than that. It is investing in technology that is upgradeable in capacity and performance, portable from one topology to another, compatible with any operating system, and fully user serviceable. In other words total cost of ownership and lifetime value. If you want investment protection from your storage solution, work with a vendor who understands lifetime value and includes this in their discussion of investment protection. If it is simply a discussion of ROI, they are only trying to justify the expense of the purchase, not the investment protection. Step 7. Ensure Reliability Without The Complexity Many of the individual components of a storage system – particularly the disk array – are extremely reliable. For instance, the RAID-configured, hot-swappable drives are each rated for 1,000,000 hours MTBF – about 115 years. The dual redundant hot-swappable controllers are each rated at 500,000 hours – about 57 years. So are the dual redundant hot-swappable power supplies and dual redundant hot-swappable fans. Even the controller cache has redundant backup batteries. What many people forget is that storage architecture failures are inversely proportional to the number of components. In essence, the more components you have, the lower the overall MTBF. For instance, if the system has 500 components (drives, switches, controllers, fans, power supplies, etc.) each with 1,000,000 hour MTBF, the actual MTBF of the entire system is only 2,000 hours – which is less than three months. There are other reasons as well. A recent Gartner Group report indicates that 80% of application downtime is a result of human and process problems, and only 20% as a result of technology failure, environmental failure or disaster. In addition, according to an OnTrack report, a recent customer survey indicated that customers say up to 78% of data loss was as a result of hardware or system malfunction, and only 22% as a result of human error, environmental failure or disaster. A rule of thumb in storage systems is that complexity goes hand-in-hand with unreliability. This is why designing a storage architecture is a test in simplicity. The more complex the design, the more points of failure are introduced into the system. Don’t let a complex system design lull you into a false sense of security regarding reliability. To simplify your storage system, start by demanding more of your disk array. The disk array should offer redundancy for both storage and connectivity, built-in security for access to drives and components, and it should be fully field manageable and configurable. Plus, if the disk array offers both SCSI and Fibre Channel connections, and has built-in firmware for open system compatibility, then it will provide multi-hosting, support multiple networks, and optimize disk usage without adding layers of cost complexity and management overhead. Step 8. Plug-and-Play Isn’t Just For Windows Anymore When it comes to deploying, upgrading and managing a storage system, plug-and-play is a must. With a new breed of open systems compatible disk arrays, plug-and-play is also a reality. This means that these new hard disk arrays can expand capacity, connect to multiple networks and operating systems, and support multiple RAID configurations without using drivers, connectivity kits or software upgrades. They also do this without interfering with existing backup or management software, and generally are fully user configurable and manageable. Not only does plug-and-play simplify system design, but it also significantly reduces or eliminates the need for environmental and performance testing before production deployment. Perhaps most importantly, it minimizes resources needed to deploy and manage the system – a critical criteria for selecting storage systems in today’s resource strapped IT department. So if simplicity and improved manageability are important to you, make plug-and-play a key part of your storage system selection criteria. Step 9. Get Your Money’s Worth Some industry “experts” like to downplay Direct-Attached Storage as a viable storage alternative to SAN and NAS. They claim that although DAS is tempting, many companies are underutilizing the systems, and may only use half of the storage. Hence they recommend spending 20 cents per megabyte on complex software to manage the 8 cents per megabyte worth of storage that is underutilized. How exactly does that save money or make the IT resources more productive? With the continuing surge in storage requirements, how can someone knowledgeable in the storage industry tell you with a straight face that the extra storage you have lying around won’t be utilized? I tend to agree with reports from Forrester Research who estimate that, on average, business capacity needs are increasing at a rate of 52% per year. So, by the time you design and deploy a Storage Area Management system to virtualize this underutilized storage, it will be long gone. The value proposition to purchase Storage Resource Management and Storage Area Management tools shouldn’t be about reallocating available current storage, it should be about helping you predict and maximize storage growth and performance. If you want to get your money’s worth from your storage system, focus on the disk array. Select a disk array to scale your capacity and set your performance expectations. Then, as the costs for storage continue to drop, and the costs to manage storage continue to go up, you’ll have a lot more money left at the end of the day to spend on your business rather than on your data. Step 10. Don’t Lose Sight of The Real Problem Like you, I have read a lot of information on DAS, SAN, NAS, Fibre Channel, iSCSI, Infiniband, Virtualization, etc. over the past few months. Not much of it seems to make the decision of which technology to use any easier or less complex. This is because most of what is available to read on this subject focuses on the problems and issues created by adding complexity between the server and the disk array. When you look for the simpler solution, you find that the disk array itself can solve or eliminate many of the issues and problems that much of the new technology is trying to address, such as compatibility, redundancy, multi-hosting, scalability and manageability. Conclusions Understanding, deploying and managing storage solutions is not as complex as it is made out to be. The objective is to simplify the process of centralizing, protecting and managing company data, i.e. streamlining the process to get the data from the server to the disk array. So before you put too much effort creating complexity between the user and the disk array, put more stock in the disk array itself. After all, the disk array is where you store your most precious asset – your company’s information. Today’s performance disk arrays may offer you far simpler and more cost effective ways to create fast, reliable, secure, scalable and manageable storage solutions, while still providing you a smarter path to investment protection.
contact us | site map | privacy |
|
|
|
|||||||||||||||
PDF
Version
