Cloud Computing

The #Covid-19 is going to be around for some more time, but the world will go on and is just one of the several cosmic manifestations in this universe. As humans, we have a limitation of viewing these cosmic manifestations of universe in a 3 dimensional view. A 4th dimension, immediately takes us to an unimaginable infinite loop of complex visualisations like a tesseract, the 4th dimension we are talking about is nothing, but "TIME". Let's hope that this 4th dimension will definitely help us coming out of this #Covid19 crisis too.

The maximum gravity we can experience is that of our earth, and possibility of us doing a time travel to look at what is in our future is near to impossible. (A #tachyon particle is a hypothetical particle that always travels faster than light), but with some simple work and very simple #linear #regression models, we can always predict or forecast a scenario to a small extent (Provided, we have enough input data & output data with a #correlation co-efficient between them close to 1).

We tried apply the 4th dimension to Cloud adoption strategy and came up with the real world use case for IT infrastructure and arrived at below calculations.

Use Case: A company plans to create a robust IT infrastructure roadmap and want to make a decision on whether to continue their existing on-premise infrastructure or adopt a cloud strategy and the total capacity required for the IT infrastructure was -

• Let's say they require 1000 VMs with the average VM being:
  • 2 vCPUs
  • 8 GB memory
  • 200 GB storage @ 300 IOPS
• Hence we need:
  • 2000 vCPUs
  • 8 TB memory
  • 200 TB @ 300,000 IOPS

How we did a 3-year TCO comparison of Cumulus™ with other private/hybrid cloud options

3 Year total cost of ownership when using public clouds

How the licensing of cloud matters when infrastructure at scale

Note:- Cloud license comparison for only non-HCI models (HW cost excluded)
(No of nodes = baremetal servers)

• Canonical - Perpetual license (Upfront cost is high, Linear cost at scale)
• Pinaka - Perpetual license (Upfront cost is low, Linear cost at scale)
• VMware - License per CPU sockets (Upfront cost is high & cost as multiples of socket pair at scale)
• Redhat - License per CPU sockets (Upfront cost is high & cost as multiples of socket pair scale)

More info - Breakdown of the TCO calculations

Assumptions, Exclusions & Exceptions in our Calculations.

• Low cost commodity infrastructure: Network, Load Balancer, Firewall, Backups
• Small, agile DevOps team to manage the infrastructure:
• Low data center costs by utilising fewer racks in an outsourced high quality data center
• Low staff costs by utilising high level of automation of small devops team and very simple architecture
• Advanced IT consolidation technologies like Docker, Containers are not included, calculations are at the level of virtual machines only.
• Typical enterprise IT infrastructure is significantly more expensive as it is based on expensive legacy "big brand" products, are inefficient and require a vast number of people to support.
• Typical enterprise infrastructure have high cost due to: Servers, Storage, Networking, Backup
• Staff: Large number of IT infrastructure staff working in many silos of IT teams including dedicated staff for procurement, architecture, facilities, networking, storage, backup, server management, VMware management. IT silos need to coordinate the activities between the teams hence are slow to provide business outcomes
• Please note that VMware cloud on AWS does not has backups service as part of their standard offering.