Too Much of a Good Thing: 3 Ways to Prioritize Infrastructure Automation

As last week’s AWS outage shows, manual configuration can be error prone.

Channel Partners

March 6, 2017

4 Min Read
Too Much of a Good Thing: 3 Ways to Prioritize Infrastructure Automation

Raju ChekuriBy Raju Chekuri

Your technical staffers have seen their jobs change in recent years. Instead of doing everything by hand, we now have tools that can handle the tedious day-to-day work of managing customer networks, freeing engineers to do more planning and strategic work for the business. After all, it doesn’t make much sense to pay high IT salaries for time-consuming and repetitive configurations, patching and the like.

Determining what should be your next automation candidate, however, is no easy task.

The logical approach is to survey the most frequently recurring problems at customer sites. Maybe you see network slowdowns, poor CPU utilization or any number of common issues affecting users or wasting money. However, automating those tasks, such as by automatically de-prioritizing low-priority traffic like media downloads, may not fix the problem.

Before throwing automation at the problem, attempt to discover the root cause. Maybe slowdowns aren’t because of end users streaming Spotify, but a lack of capacity brought on by more use of software as a service. To find out, you must analyze top consumers of bandwidth over a period of time, and then rationalize the optimal bandwidth to allocate to various traffic segments based on business priorities.

The lesson is: Search for and fix the root cause before applying tools to a problem. Why not automate everything, you may ask, and cover all the bases?

That sounds sensible in this day and age, but it’s not. If a task occurs only a couple of times per year per customer site and doesn’t entail significant manual effort, consider the cost of acquiring and managing a tool to automate the task. It probably won’t be worth the investment. A good example is server provisioning, which now happens even less frequently given the move toward cloud-based infrastructure.

When evaluating new opportunities for automation, prioritize tools that can reduce manual labor, improve the quality of the customer network and reduce defects — and, as a result, simplify complex troubleshooting so that you don’t have to put senior engineers on the job.

Categorize manual tasks into three buckets:

  1. Too involved. The process is a complex, multistep procedure involving multiple systems and subsystems. This is a prime candidate for automation, because of the complexity involved. That increases the risk of failure from human error. See: last week’s AWS outage.

  2. Too long and tedious. The process takes a long time to complete. For example, setting up a software-defined data center requires an individual to run manual tasks over several days, but not continually. So instead of having an employee monitoring a project with long periods of dead time in between work, you automate the entire process. This can also reduce errors.

  3. Neither one. The task fits into neither category, which means it doesn’t need to be automated — yet. This is an exercise you should repeat regularly.

Prioritization and Tools

Now that you have a list of automation projects, which should you take on first?

It’s useful to look at these activities within the three phases of infrastructure lifecycle management (ILM): provisioning, operations and audit. Of those areas, one likely has a higher business impact for you or your customers. A hosting company, for example, would prioritize on being most efficient in provisioning, because that drives revenue. A company in which compliance is critical to staying in business should focus on automating auditing tasks first.

Finally, you’ll have to make some decisions around tools.  A large enterprise or MSP can afford automation suites for each core infrastructure area: networking, storage and servers. A smaller company will likely need to pick and choose. Select tools that fit the infrastructure area where your most common problems occur, or where you have the most requirements.

While there are plenty of commercial tools on the market, acquiring lots of point tools gets expensive. If your team is strong technically, go the open-source route. Your team will be able to handle the extra work involved to configure and learn the tools, and you will save a lot of money. Otherwise, choose commercial tools where ease of use is built in and support is always available from reputable vendors.

Today, there are many excellent reasons to focus on IT automation. It makes operations more accurate and usually more cost-effective. It allows for the standardization of IT processes to reliably meet business needs. Yet there’s always a balance. We cannot replace all humans in the data center, and intelligent decision-making by experienced engineers will continue to be a highly valuable asset that will differentiate your services among customer CIOs and CTOs. The trick is to balance humans and machines, with a goal toward extreme efficiency and supporting business objectives.

Raju Chekuri is CEO of NetEnrich.

Read more about:

Free Newsletters for the Channel
Register for Your Free Newsletter Now

You May Also Like