The Uberization of Analytics

As I have traveled the country talking with customers, I have gotten a common refrain that companies want to move to the cloud for analytics to reduce costs. My retort is that you need to be careful and understand what you are really trying to achieve, for while there may be good reasons to run your analytics in the cloud, saving money is not a given.

What is the objective?

The analogy that I use is having a car. If I travel into a city and will only be there a few days, with much of my time in hotels or offices, it would be silly to buy a car for the effort. Instead, I take a cab or use a ride-hailing option. I only want to pay for the car, or more importantly, the service when I need to use it.

Now if I am going to have a long trip and will need a car frequently or am driving long distances, then it may make more sense to rent a car, even though I am “paying” for the vehicle to just sit in the hotel parking lot. I am paying for the convenience of having the car instantly available, as well as being able to keep things in the trunk as needed.

But when I am home and I use my car to drive to and from work each day, to ferry the kids back and forth from their activities, and would like to have a consistent driving experience, I have a different outcome in mind — and so I buy, or lease, a car.

Analytics in the Cloud

So looking at your business and analytic needs, we see a similar shift happening today. Some analytic workload demands consistency, governance, wide sharing of the data, and a 24/7 ingest and access environment. In these situations it makes sense to “own” the system, which may be either a platform within your own data center or perhaps a managed cloud environment with a long-term lease.

Other analytic workloads are temporary in nature, such as a data scientist exploring a new data set, a business user wanting to have a data lab (sandbox) to try new ideas or a database administrator looking to test a new application or ecosystem tool. These options do not warrant purchase or long-term ownership, but instead can be easily accommodated with a public or private cloud option that can quickly be created, used, then dropped.

When to use which is the question

Clearly there are options available and, when making the decision, one must understand the value — and the cost — before proceeding. For example, I can use Uber or Lyft as my everyday transport even though it may be more expensive than owning a car. The benefit is that I would no longer have insurance, maintenance, gas, the need for a garage, and so on to worry about. Thus the tradeoff may be worth it when everything is considered from both a cost and convenience perspective.

In the cloud, you have the tradeoff from many angles: cost, system management, data transfer, frequency of access, privacy, security, and availability, to name a few.

Just like with your transportation options, there will most likely be a mixture of the use cases throughout your analytical ecosystem, and knowing how to leverage each deployment alternative appropriately — using the right tool for the job — is the ultimate objective.

That’s all for now because I need to get going — my ride just arrived!