Despite the image it conveys, the cloud consists of a physical infrastructure: many computers housed in massive data centers all over the world. Various hyperscale cloud service providers are competing on service quality and pricing to offer virtualized cloud services, but they are facing an increasing number of cloud challenges: building faster, denser, bigger, cost-effective and more power-efficient data center infrastructure that can be scalable, sustainable and resilient.

The Statistics Impacting Cloud Challenges

data center trends graphic

According to the latest Cisco Visual Networking Index (VNI) forecast for 2020, global IP traffic will reach 2.3 zettabytes (ZB), a 100-fold increase from 2005. Smartphone traffic will exceed PC traffic for the first time, and more than 23 billion IP-enabled devices – also known as Internet of Things (IoT) devices – will be connected to the network.

According to the Gartner Worldwide IT Spending Forecast, for every 200 additional mobile devices added to a network, one more server in the cloud will be required for data processing. Increasing numbers of devices is the source of many cloud challenges.

Today, nearly 75% of data center traffic remains within data centers for computing, processing and storage. For example, Facebook users are uploading more than 300 million pictures per day; each of these pictures is stored in different locations in different format sizes for caching and reviewing. As more and more pictures are uploaded, cloud challenges increase.

High-performance computing, on the other hand, has increased the maximum FLOPS (floating-point operations per second – an indicator of computer performance) by a factor of two annually, and the No. 1  supercomputer from Sunway TaihuLight in China has already reached 93 petaFLOPS (93×1015), with more than 10 million CPUs in 40,960 computing modules.

Cloud migration is no longer an emerging trend, but an ongoing action that takes place in most organizations to meet booming bandwidth and computing/storage capacity needs created by IoT, industrial data processing and central office IT services, including onsite private cloud services and offsite public cloud services.

9 Tips to Improve System UptimeNew Cloud Challenges

As cloud leaders, such as Google, Amazon, Microsoft and Facebook, are investing heavily in technology innovation, designing new topologies to best suit their needs and pushing hardware to customize the most cost-efficient components through multi-source agreements, the industrial standards bodies – IEEE, TIA and ISO/IEC – are accelerating to develop and release new standards that support this growth.

chart listing the CAPEX spending of the top 6 cloud owners

Source: Synergy Research Group

There are some fundamental cloud challenges, however, that cannot be solved by hyperscale data centers:

  1. IoT needs to work even when the connection is temporally unavailable.
  2. Cloud data centers cannot provision enough bandwidth or appropriate service levels to cover every corner of the world.
  3. Centralized data center architecture cannot meet the emerging requirements of instantaneous data analysis due to the latency of data generation, transmission and processing.
  4. It’s not viable for cloud service providers to optimize infrastructure costs by only providing services from their own data centers.

Solving Cloud Challenges

graphic listing high speed migration challenges and signal transmission technologies

Can we solve these cloud challenges by building more hyperscale data centers? More hyperscale data centers would be great, but that’s not the answer to everything. The return on investment would diminish for further scale increase due to the premise and infrastructure costs:

  1. Fixed costs have already been amortized across very large number of servers.
  2. Migration from hard disk to flash storage will reduce the processing latency considerably, but would further increase the cluster bandwidth.
  3. Power consumption limits geographical data center locations.

Fortunately, we already have some new solutions to cloud challenges on the horizon:

  1. Distributed data center topology, which allows for quick disaster recovery. Data center interconnect (DCI) solutions have been developed to efficiently connect geographically separated data centers in the same region.
  2. Edge computing platforms that support quick data analytics locally can address the latency-sensitive applications in the new ecosystem of IoT, such as connected autonomous vehicles.
  3. Multi-tenant data centers (MTDCs), as a fast-growing segment, provide colocation services such as hosting the private cloud infrastructure and providing public cloud access to end users. They can help enterprise customers reduce cost of ownership, allowing the content delivery network and cloud service providers to quickly expand their service coverage.

The evolution of the information era will continue, and new cloud challenges will emerge, but one thing is for sure: We can certainly expect a bigger and better cloud in the new, connected world.

Are you currently struggling with any of the cloud challenges we mentioned in this blog?
Share your experience with us in the comments section below!