Ten Key Trends in Storage Administration - Page 2
#5 Storage Tiers
Storage tiers have been around for decades and used to be called HSM; same function, different name today. The cost of really fast storage (SSDs) in enterprise storage is still five to eight times greater than the cost of spinning disk, but everyone wants SSD performance at disk cost (we all want a Porsche for the price of a Chevy, too). The functionality of tiered storage is reasonably easy, but the hard part is how to manage important applications, as well as less important applications and applications that you do not care about at all within a single namespace and still get the right results out of your tiering software. Management of storage tiers within a workflow is difficult, especially for complex changing workloads, and determining what is important and what is not is also very difficult. A single misbehaving application can hurt the performance of critical work.
#4 Integrating Old and New Technology
How you deal with old technology will have a lot to do with how successful you are in the organization over the long haul. The integration costs of mixing old and new technology can be very high, but the costs of doing nothing will be higher. Figuring out how to integrate old and new technology into the same environment is as much art as science, and doing all of this while meeting your business objectives is not only critical for the organization, but critical for your success in the organization.
#3 New Technologies
Picking the right new technologies is difficult, as you are betting on winners and losers. For a recent example, let’s say you picked an SSD vendor that provided you with a large controller five years ago. There is a high statistical probability that that vendor has been consolidated or is no longer in business. The SSD example I used is similar to other vendor trends that have come and gone. It seems to happen every ten years or so. Anyone remember the large number of new storage controller vendors of the early 2000s, and all the new compute vendors in the early 1990s (Kendall Square, MultiFlow etc.)? Picking the right vendors for new technology requires a good understanding of both funding and technology.
#2 The Cloud
Not a day goes by where you don’t hear a prediction that a certain percentage of all computing will be moving to the cloud by a date in the not too distant future. If all of this is true, many of us are going to be out of jobs. Like it or not, many of our organization’s business functions are going to move to the cloud, and fighting tooth and nail for each application is not likely a good long-term strategy to ensure continued employment. The key here is to figure out what common activities can and should move to the cloud, with a primary example being email, and what applications should not move to the cloud. You are going to have to develop cost models and issues for the long term to show what makes the right business sense. Just saying no is not the right answer.
The quickest way to lose your job is to have a security breach with your architecture or your technical decisions. Security needs to be the number one thing you think about all of the time. Of course, many security breaches will not be in the storage domain, but it is likely that old saying about throwing the baby out with the bathwater might apply. Have a documentation trail showing that you have done your best to ensure that the system is as safe as you can make it.
This isn’t an easy list to stay on top of, but will be important to your continued success in our fast-changing industry. This list is a snapshot of what I see today and is likely good for at least a few years. For sure, this is not a static list — just like we are not in a static industry — but things in our industry seem to be cyclical. I have long said that there are no new engineering problems, just new engineers solving old problems.
Photo courtesy of Shutterstock.