Software robots are taking over the data center!


Image: Jeff Sandquist (cc:by)

By now, you’re probably familiar with IT’s new business mindset: that of a highly efficient service provider. Cloud services, consumerization and BYOD are just a few of the business-driven topics that are currently generating buzz.

However, another major trend has begun to silently transform the data center by providing IT equipment with an intelligence all its own. It turns out that, in order to truly modernize IT, your application servers, networks, data storage systems, and even end-user devices must live intelligently—often, independently.

What do I mean by that? Read on...

Successful cloud service providers offer some good examples to follow. Such providers rely on increasingly automated components in their cloud-based infrastructures. For them, automation occurs at multiple levels, allowing for constant provisioning and monitoring of compute, network and storage resources.

With so many moving parts to track and a growing mountain of data transactions to process, service providers utilize software-based automation to abstract and simplify the complex operations of underlying data operations. On a smaller scale, traditional IT shops are also beginning to utilize these sophisticated algorithms to automate many former, labor-intensive processes.

  • Note: You may think that your IT department is already automated, via virtual servers with their portable, policy-based workloads. In reality, these are just the tip of the automation iceberg.

Data-center automation in a modern IT world

I tend to think of automation as the linking of IT equipment to software in such a way that the equipment becomes self-acting or self-regulating. Today’s IT infrastructure requires increasingly sophisticated and intelligent systems that not only anticipate user needs but also work together to simplify and automate many common functions.

A couple of examples come to mind: mobile devices that are auto-discovered/auto-configured and network switches that load balance by automatically routing data packets to avoid bottlenecks.

While this type of automation has been commonplace in servers and networks for years, it has been slow to come to the world of data storage. This is changing, however, for two important reasons:

  1. The range of options available in enterprise storage arrays has become mind-boggling, and
  2. Tools have become available to take mundane, repetitive task from the hands of error-prone storage admins and place them into software-based policy engines (i.e., software robots.)

Automation comes to data storage

The storage array controller has evolved from its early, mainframe days of “dumb” bus and tag communications into a highly sophisticated data server. While most enterprise arrays have considerably more features and options than my example below, I’ll use it to show how things can get complicated rather quickly.

Let’s say you have a basic storage array with just 15 adjustable storage features. We’ll also assume that each storage feature has just three basic configuration options -minimal, moderate and maximum. The following table shows one example of how you might choose to configure the system:

sample storage features

Basic Storage Configuration Example

Perhaps you pass this chart around to your fellow admins, and each suggests a slight change. No problem. After all, how many possible combinations of selections could be possible here? 45? 225?

Believe it or not, in this table alone, there are over 14 million possible combinations of check boxes! That’s based on just 15 selections, each with only 3 configuration choices. (For those of us who were not math majors, the exact number of possible combinations is 14,348,907 or 315.)

Today’s infrastructure requires automation

To make matters worse, instead of just 15 features with three configuration options each, a modern storage array actually has dozens (if not hundreds) of tunable parameters, each with dozens of possible settings. It’s doubtful that any agreement would ever be reached between admins—and, as we all know, things change quickly in IT. Thus, the configuration matrix would most likely require constant tweaking.

With so many tunable variables, there can also be any number of potential, negative consequences stemming from an improperly-tuned storage array, ranging from unhappy users to dreaded, out-of-space conditions. This is why storage admins are relying more and more on automated policy engines.

Are you ready for the future?

The days are quickly ending when a storage administrator could tinker a little bit with his/her storage to get it properly provisioned. Old storage tricks like short-stroking, wide-striping and over-provisioning just won’t hack it anymore.

How much time are you still spending manually tuning individual arrays versus investigating how they might be automated within the emerging paradigm of the software-defined data center (SDDC)?

Are you ready to embrace the role of an IT service provider? If so, it’s time to change your focus away from managing painstaking manual tasks. Instead, succumb to the growing world of IT automation.

Computerworld's IT Salary Survey 2017 results
Shop Tech Products at Amazon