Data Gravity and Cloud Computing

Cloud computing is growing rapidly as a deployment platform for IT infrastructure because it can offer significant benefits. But cloud computing is not always the answer, nor will it replace all of our on-prem computing systems anytime soon—no matter what the pundits are saying.

It is true, though, that moving some types of workload into the cloud can possibly lower operating costs. Establishing and running a data center can be very expensive. Shifting to the cloud offloads at least some of that cost. Deploying in the cloud can also offer higher flexibility and easier scalability.

These, and other issues, have caused many to believe that cloud computing will dominate all of IT in a short period of time. But this is typical industry hype. My long-standing opinion is that although cloud computing will continue to grow, on-prem computing will not be obsolete anytime soon.

And there is some recent evidence to back up this opinion.

TechTarget’s “2020 IT Priorities Survey” indicated that on-prem data centers remain the dominant infrastructure deployment platform, with 45% citing on-prem data centers as their primary computing platform. Additionally, analysts at IDC indicate that 86% of enterprises are thinking about or actually repatriating cloud workloads back on-prem. If the cloud is going to quickly overtake all of computing, why do we see such results as these in 2020, nearly 15 years after the introduction of the term “cloud computing?”

Well, I believe there to be several compelling reasons. At the top of the list is a capacity issue. Consider of all the data centers “out there” running production workloads for small-, medium-, and large-sized businesses. Most organizations have a plethora of Unix, Linux, and Windows servers backed up by terabytes of storage. And that does not even consider the mainframe. Now let’s shut them all off and move the workloads to cloud service providers … Do the CSPs have the capacity to support that amount of work? Certainly not! OK, then, how long will it take them to build it out? Is there enough bandwidth for all of that workload? 

Which brings me to the next compelling issue: bandwidth versus data gravity. These two growing, rival forces are the biggest factors influencing IT today.

Data continues to grow unabated. IDC predicts the total size of what it calls the “Global Datasphere” will grow to 175 zettabytes by 2025. A zettabyte is 2 to the 70th power bytes, or 1 followed by 21 zeros. One zettabyte is approximately equal to 1 billion terabytes. That’s a lot of data!

But what do I mean by data gravity? Data gravity is the concept that data remains where it is and applications and services (and even more data) are attracted to and use that data.

Moving data can be a tricky, time-consuming, and complex process, even on-prem. Now, add in the movement of large quantities of enterprise data through the internet to a cloud service provider. Anyone who has managed or used big data, or even moderately large data, will immediately understand the problem.

But, while data continues to expand, and data gravity causes it to stay put, bandwidth also grows. Nielsen’s Law of Internet Bandwidth proposes that users’ bandwidth grows by 50% per year. So, bandwidth gets an order of magnitude faster about every 5 years. This means that the rate of bandwidth growth is faster than the rate of data growth but, unlike bandwidth, there is really no practical physical limit on data growth.

The competing forces of data gravity and bandwidth will continue to impact the adoption and success of cloud computing for the foreseeable future. As the amount of data continues to grow, and stick where it lands, organizations will have a difficult time eliminating all of their on-prem computing infrastructure. So, hybrid multi-cloud scenarios, with a combination of on-prem workloads and multiple cloud computing services, will likely dominate the IT landscape for the upcoming era of computing.

Share this post

Craig Mullins

Craig Mullins

Craig S. Mullins is a data management strategist and principal consultant for Mullins Consulting, Inc. He has three decades of experience in the field of database management, including working with DB2 for z/OS since Version 1. Craig is also an IBM Information Champion and is the author of two books: DB2 Developer’s Guide and Database Administration:The Complete Guide to Practices and Procedures. You can contact Craig via his website.

scroll to top