Every year Cisco produces a Global Cloud Index, a report that was developed to estimate (and it is just an estimate) global data center traffic growth and general trends. The report is a complementary resource to Cisco’s more general IP network studies, but it provides more meat for which cloud-specific pundits can chew on.
+ Also on Network World: Enterprise IT pros see most workloads in cloud by 2018 +
This year’s report took a look at application workloads, highlighting the number of global data centers in what Cisco classifies as hyperscale-size and compared total storage capacity within data centers versus what is actually being stored. This last metric is important because, as Gavin Belson so correctly pointed out in the HBO parody series Silicon Valley, the amount of data being produced and stored is greatly increasing—providers need to think about capacity. While the show is very much tongue in cheek, it has often been the case that it closely mirrors the real world:
“Data creation is exploding. With all the selfies and useless files people refuse to delete on the cloud, was created in the last two years alone. At the current rate, the world's data storage capacity will be overtaken by next spring. It will be nothing short of a catastrophe. Data shortages, data rationing, data black markets. Someone's compression will save the world from data-geddon, and it sure as hell better be Nucleus and not goddamn Pied Piper! I don't know about you people, but I don't wanna live in a world where someone else makes the world a better place better than we do.”
Anyway, back to the real world and Cisco’s study. At a high level, cloud traffic is expected to rise 3.7-fold, up from 3.9 zettabytes (ZB) per year in 2015 to 14.1 ZB per year by 2020. This rapid growth of cloud traffic is attributed to increased migration to cloud architectures due to their ability to scale quickly and efficiently support more workloads than traditional data centers.
Workloads growing
The interesting thing is the increase in the adoption of both virtualization and even newer technologies such as containerization, which tend to deliver greater operational efficiencies for data center operators. This somewhat reduces the impact of this data and application growth in terms of a pressure on capacity. So, to the numbers:
- Business workloads dominate data center applications and are growing. The study found that business workloads will grow by 2.4-fold from 2015 to 2020, but their overall share of data center workloads will decrease from 79 percent to 72 percent.
- Consumer workloads, while smaller in number, are growing faster. Perhaps unsurprisingly, during the same time, consumer workloads will grow faster by 3.5-fold. By 2020, consumer workloads will account for 28 percent (134.3 million) of total data center workloads, compared to 21 percent (38.6 million) in 2015.
- Internet of Things/analytics/database workloads are growing the most in terms of share of business workloads, with collaboration and compute workloads largely maintaining their share. By 2020, database/analytics/IoT workloads will account for 22 percent of total business workloads, compared to 20 percent in 2015.
- Video and social networking will lead the increase in consumer workloads, each respectively grows their percentage significantly. By 2020, video streaming workloads will account for 34 percent of total consumer workloads, compared to 29 percent in 2015; social networking workloads will account for 24 percent of total consumer workloads, compared to 20 percent in 2015; and search workloads will account for 15 percent of total consumer workloads, compared to 17 percent in 2015.
In terms of the quantity of hyperscale data centers, the report suggests that the 259 facilities Cisco classes as hyperscale globally today will grow to 485 by 2020. This is a serious amount of capacity (not to mention power, land usage, air conditioning and other infrastructure to build out—a boon for specialist data center builders and operators). By 2020, Cisco predicts these hyperscale data centers will surpass 50 percent of all data center traffic, somewhat at odds with the “computing at the edge” approach that many people have predicted.
There are some interesting, more detailed findings related to networking traffic, the rise of software-defined networking (SDN) and the growth of public cloud versus private cloud. All detailed reading for those who are interested, but fundamentally, this report can be summed up simply:
The amount of traffic, data and compute is trending upwards sharply. This upwards trend is going to require more data center capacity and more data centers to be built at mass scale. While in theory we could simply see an increased buildout in the size and style of data centers that we’re used to, the increasing economic pressure will see a huge amount of investment in new technologies to build data center efficiencies—from server design to air conditioning, from power generation to water recycling. It’s going to be an interesting decade or so for data center operators.