Alien Invasion – Hyperscale

What are the network effects operating at hyperscale?

The Alien Invaders in the highest tier of cloud and computing infrastructure business landscapes include Amazon Web Services, IBM Cloud, Google Cloud, Microsoft Azure, and Alibaba Cloud. The next tier includes HP Cloud, Oracle Cloud, and enterprise providers “riding along” such as GM Cloud, Ford Cloud, and many pre-existing data center providers (Verizon Business, NTT, iPower, GoDaddy, etc.). The hyperscale competitors participating in this market have distinct advantages even for the networks operated by huge competitors on a global scale.

EkaLore will explore the “electricity to EPS” style of calculated capital efficiency in an enterprise in a separate BOTE (Back of the Envelope) post. This post’s focus lists strategic advantages.

The Hyperscale Cloud competitors each possess certain characteristics:

1)    Platform Agnostic Characters – AWS can apply gigantic scale GPU compute, IBM Mainframe GIPS, and multiple scales of hyper-compute clusters. Operating systems, network environments, and operations management infrastructures can all “co-exist” or be supplanted by Amazon Cloud operations. Microsoft Azure has similar approaches to supporting highly diverse enterprises desirous of deploying into the Microsoft Azure infrastructure of data centers, networking, and operations management.

2)    Tool Agnostic Characters – from hardware shared GPU/desktops (such as those deployed by VMWare, Citrix, or others), to exotic software tools for highly specialized AI, Big Data, or Data Sciences the Hyperscale Cloud competitors have “a way” (not always equal or efficient) or providing the tools desired by customers and connections to make them effective

3)    Gigantic resources on demand—capacity at a price. Hyperscale competitors offer compute, storage, network bandwidth and I/O processing that was unimaginable in the past. Maybe more importantly on demand. Compute offerings include (TFlops or GPU), gigantic clusters of CPUs (or various makes and models). Storage offerings go up to Petabytes (on demand), network bandwidth (connections limited usually by capacity in destination connected nodes), and I/O processing capacity (for I/O intensive problems – such as the volumes of data from multiple daily scans of the entire planet from satellites).

Historically, HP, IBM, and other companies offered dynamic capacity. The new top tier hyperscale players have outstripped these old-line computing compaines in the ability to scale up and scale wide dynamically.

4)    Price choices – opaque and transparent. Large scale deployments (especially of on-demand capacity and capability) are difficult to estimate even using the hyper scale competitors' tools. In many cases (storage, redundancy) the inherent character of cloud processing provides capabilities over ordinary commercial “shop deployments”. The prices for compute chores (such as Zillow’s repricing of all homes in the US for ‘ZEstimates”) may vary from service to service while the key is understanding there are choices to operating this on multiple services (scale out wide or scale up the intensity).

5)    Security – simply the dynamic security environments presented by Cloud Vendors exceed the capability and talents found even in good size enterprise production environments. The ability of staff talent is huge to access tools against environmental threats like DDoS, Ransomware, brute-force crypto/operations attacks, or network penetration analysis (using deep packet inspection or other network management techniques). Whereas even a large scale enterprise has limited ability to access and operate sophisticated tool environments the hyperscale cloud competitors use hyperscale and tiers of tool environments routinely to defend their cloud data centers and network provisioning. Alone, the savings from the availability, allocation, and operations of comprehensive intrusion detection, consistent crypto handling, and network threat management (such as exfiltration controls) is greater than that available to SME enterprises. These costs are “baked in” (understanding that different vulnerabilities come with the processes and methods) from other contemporary data center deployments.

These are only some of the reasons why “colossal gets bigger” sees the largest hyperscale competitors growing even larger.

EkaLore helps companies recognize strategic advantage in their environment and take advantage of it.

Learn more about our approach at www.ekalore.com/alien-invaders