Slideshow Avoiding 6 Big Data Storage Management Issues

Published
  • September 27 2015, 8:49am EDT
8 Images Total

Avoiding 6 Big Data Storage Management Issues

As big data and machine learning take hold, storage technology often becomes automated, virtualized and more complex. Some platforms tend to pin their owners into a commodity lifecycle to keep pace with scalability, speeds and costs. Here's how to break out of the commodity storage lifecycle.

Step 1: Break the Commodity Cycle

When you open a bottle of wine, you give yourself two choices – drink it soon, or watch it go bad. The storage lifecycle is similar. Even if your business doesn’t consume the entirety of a storage purchase, it will eventually need to be replaced to continue providing benefits to your IT system. By leveraging services and public cloud options, you can consume storage capacity as you need it and avoid watching your original purchase go stale.

Content Continues Below


Step 2: Achieve Balance Between Fast and Slow

Increasing speeds and performance won’t necessarily improve your user experience. If you’re attuned to your various systems’ needs, you know that different data sets run at different intensities and can therefore run optimally without the fastest and most expensive data storage. Sure, some workloads should be assigned exclusively to your fastest tiers, but even seasoned IT pros can overestimate the amount of high-performance data they actually need. Don’t be afraid to gauge end-user needs and relegate less performance-sensitive data to colder tiers of storage. In turn, this practice will justify the costs of high-performance storage.

Step 3: Keep Your Data On The Move

Although your data gets colder over time, keep in mind that it will probably need to be pulled back to high-performance areas of your stack at some point in the future. To stop performance bottlenecks, avoid storage systems that incorrectly estimate your data usage, or are too slow to react to changes in performance needs as your data goes through its lifecycle. Instead, your IT environment should anticipate and accommodate frequent changes in data temperature and location to support performance and cut storage latency.

Step 4: Prepare to Navigate a Lopsided Price Curve

With the huge price disparity between high performance, high endurance flash and slow-speed spinning media, what’s the right dollar amount for you to spend on various storage tiers? The answer to this question isn’t universal – it depends entirely on the patterns and user needs associated with your data. Flash prices have become more manageable recently, but the cost of a gigabyte of flash remains orders of magnitude more expensive than the cost of disk storage. To navigate the price curve, stay cognizant of where your dollars are headed and how directly they support your capacity and scaling needs.

Content Continues Below


Step 5: Don’t Ignore the Hidden Killer of Cloud Storage

Security and control issues are frequently cited as roadblocks on the path to enterprise public cloud adoption, but, in many cases, failed cloud strategies are the handiwork of a silent killer – latency. Many companies keep compute workloads in-house and turn to public clouds for data storage, but when data moves between cloud infrastructures and end users across major distances through complex networks, serious latency issues can form. In fact, IDC reports that up to 50 percent of cloud customers have moved workloads back on-premise or to another service provider due to such issues.

Step 6: Be Ready to Scale Your Data – and Your Business

If you have the chance to expand your business significantly, storage is the last thing that should hold you back. Even as your data stores grow and new business application initiatives take hold, if your storage environment is properly designed, scaling up will not disproportionately increase IT costs. When you add new technologies to your stack, you should ensure they’re ready to grow with your business. For example, your data management policies should match the needs of your applications with the on-demand expectations of your end users, and they should bring simplicity to the equation whenever possible.

Thank You and More

This slide show was compiled by Lazarus Vekiarides, chief technology officer and co-founder of ClearSky Data, a startup focused on a new enterprise infrastructure solution. Previously, Vekiarides was a member of the core leadership team at EqualLogic and an executive at Dell. He is an expert in data storage, virtualization and networking technologies. The views in this slide show are those of Vekiarides. For more Information Management galleries please visit here.