By Frank Nishimori, Solution Architect, and Brett Anderson, Senior Director, HP Solutions, Logicalis US
When you talk to people about storage, the buzz today is all about flash. And there’s a good reason for that – flash is fast. Really fast. Most IT pros know that flash offers significant performance advantages over disk and certainly over tape, but for many, the idea of flash was out of reach due to its hefty price tag. If you read flash vendors’ marketing materials today, however, you’ll quickly get the idea that the cost of flash has come down and that it may even be comparable to spinning disk. While technically true, there is a caveat – and it’s an important one.
To bring the price of flash arrays down, one of the techniques manufacturers are using is something called “compaction” – similar to deduplication or compression of data – which essentially gives you the ability to pack more data onto a drive. The problem is, not all data sets are compressible – and that’s when CIOs who have quickly made the move to an all-flash array have an “oh-no” moment.
If you have data – jpegs, video or encrypted data, for example – that cannot be further compressed, then you’re going to need a scalable solution where cost-effective hard disks work hand-in-hand with flash technology to give you a single bundled solution that can store any kind of data quickly and easily.
One of the most important things to remember about flash is that the real advantage of flash isn’t speed – it’s time. The performance gains you can achieve with a flash array put valuable data into decision-makers’ hands in record time. While flash is truly fast, however, it’s not a panacea for all time-to-data issues, and all flash arrays are not created equal, which means there are a number of factors to consider before deciding if flash is right for you.
If you need faster input-output (I/O), you may find a flash array an attractive idea, but one of the chief considerations in terms of delivering that performance is whether the array you select is built for the enterprise environment. Before making your selection, ask yourself: Are the replication capabilities there? How’s the quality of service and reliability? Can you segment production and development environments? Does the solution offer self-balancing functionality? Don’t automatically assume you’ll get the same level of enterprise capability you’re accustomed to in every flash array. Look for an I/O monster with enterprise functionality that can store slow or fat data as well as provide the speed you need for business-critical, compressible data all within a single storage frame and all managed from a single pane of glass.
While flash is currently more expensive than disk, compaction techniques can result in anywhere from 4:1 to 20:1 storage capacity gains; this technique allows vendors to market flash at the same price point as disk with many times the performance. CIOs, however, are finding that if their data is already compressed, they aren’t reaping those kinds of capacity savings. The bottom line is, flash can be as inexpensive as disk storage when compaction techniques are employed, but it all depends on the kind of data being stored. Take stock of how much of your data is already compressed before counting on the cost of flash equaling that of spinning disk.
When planning to deploy an all-flash array, someone will need to be trained to manage it. And if you cannot take full advantage of flash capabilities, or if you have archival data that can be more cost-effectively stored in other ways, it could take multiple people just to manage your organization’s varied storage environments. Look instead for an array that combines the performance advantages of flash with the cost advantages of a general-purpose array in a single frame with an easy-to-use interface. The easier a storage environment is to manage, the fewer people will be needed for this task. The ultimate goal: Select a solution that requires you to dedicate only part of a single person’s time to managing the organization’s storage environment.
When planning a migration of corporate data to a flash array, consider the timing. Is it better to do this over a weekend when the entire IT infrastructure can be shut down? Or can you perform a live migration without outages of active data? If the flash array has the ability to pull data from the former environment while requests are being made, a mid-week transfer may be possible, but this is the exception rather than the rule, so it’s critical to determine what kind of downtime may be required in advance of migration day.
What happens when your organization grows and more storage capacity is needed? Generally, flash storage is sold in vendor-defined flash modules. These modules typically come in groups of 10 to 12 drives, a costly upgrade to purchase all at once. Look instead for solutions that offer smaller upgrade increments – and be sure to ask about the controller limits as well. It’s not all about capacity; think about I/O too. Because flash is so fast, it can cause you to exceed your storage system’s performance capabilities; if you exceed your I/O requirements, what’s the next step? Asking the right questions up front will help you select a storage solution that will not only meet today’s expectations, but one that will provide a pathway for future growth.
Want to learn more? Start by reading a Logicalis US white paper, “Changing the Speed of Business,” and downloading a flash storage infographic to learn what flash can do for you. Then, ask yourself: Is your data a liability or an asset? You can take a Logicalis US quiz to find out, then read about six ways flash is changing the storage landscape. You probably already know that your legacy storage wasn’t designed for today’s uses; click here to explore a few of the storage innovations that can solve your organization’s most demanding storage problems.