5 Step Cloud Cost Assessment & Optimization Framework

Background/Context

The last decade has seen the usage of Cloud Computing rising exponentially. Enterprises have slowly moved most of their IT infrastructure to the cloud to take genuine advantages that today’s Cloud Platforms like Azure, AWS, and GCP offer.

One of the main advantages of using Cloud Computing is moving from the Capex Heavy model of IT Infrastructure to Pay-as-you-go, On-demand and Subscription-based Opex model with the added advantage of the ease of administration and management of the Cloud resources without the liability of managing them fully like in an enterprise owned datacentre.

There are literally hundreds of services offered by the Cloud providers today and enterprises end up using a large combination of them as part of their IT workloads either developed or bought by them.

Today, Cloud Computing has become like Utilities (Power, Water, Gas) which Enterprises can’t live without, but at the same time they are realising that somewhere in building the ‘Switches, Taps & Valves’ of the Cloud Computing services are left ‘Open!’ and month on month the consumption and the bills are increasing without much central control.

Challenges & Opportunities

Enterprises can become complex organizations as they grow organically or through M&A activity. Silos are created under different departments, business units, subsidiaries to ease the management of this growth as per their mandated KRAs.

While there remain some common IT functions, like say email, which is common across the enterprise, many other IT functions are custom requirements of these different silos inside the enterprise.

Today many enterprises have CIO-CTO IT teams that have a specific mandate of centrally governing the various spread out IT functions of the enterprise. These IT teams are looked at as enablers and major supports for the various Business functions the enterprise has.

Coming back to Cloud Computing, the Opex model of billing enables these IT teams to very quickly provision IT/Cloud services to the various business functions and Silos without a large hassle of getting requisite approvals as the initial running costs are comparatively very low compared to buying the similar IT Infrastructure.

To enable this Opex model of working, Cloud providers have developed very complex metering systems for different kinds of cloud services. The metering can be as simple as a time period-based SaaS plan offering or as complex as 3-4 different measurable units getting calculated to be billed.

Interesting trivia: Before becoming the CEO of Microsoft, Satya Nadella was leading the development of the complex metering & billing system of Microsoft’s Cloud Platform Azure, then known as Windows Azure.

Due to the high complexity of the metering and billing of many important cloud services like Data Warehouse Storage & Analytics Processing, it becomes very challenging to budget and forecasts the expenses of the said architecture or solution.

Traditionally architectures of IT solutions did not consider billing as an aspect of the architecture so a lot of legacy solutions that have moved to Cloud over the years are not designed and Optimized to take advantage of the On-demand, Opex nature of the Cloud Billing. In fact, traditionally the solutions were architected from a perspective of giving high performance while handling peak load. So, when such solutions move to the cloud, teams tend to over-provision cloud resources as per the architectural design even when peak loads are encountered only sometimes in a week or seasonally.

Bringing all the above aspects together what we typically get to see in enterprises nowadays is over-provisioned cloud resources which may go unused a lot of times, a lot of orphaned resources which remain operational and billable as teams keep on trying different Pilots, POCs, Demos, etc and IT solutions not architected well enough for the Cloud thus not optimized to take advantage of the Pay-as-you, On-Demand Opex model of billing.

The good thing here is that, Cloud Providers today give a very granular picture of the metering and billing which is quite complex to understand but with a trained set of eyes we can pinpoint which tap, switch or valve is open! in our Cloud Subscriptions.

Cloud Providers have their own set of monitoring, diagnostic tools for the metering and billing which gives us a fair idea about our usage. With APIs provided by the Cloud Providers, there are some 3rd party tool providers that may give even better analytics into the Cloud Consumption of the enterprise.

Based on long term commitments, Cloud Providers are also ready to give lower billing rates or bulk discounts, etc to make sure enterprises get locked in but get a better dollar deal compared to the rack rate.

Solution

The current on-going global crisis due to the Coronavirus Pandemic is a trigger where enterprises need to quickly plan and implement Strategic but Non-Labor Cost Reduction across the spectrum.

Cloud Cost Optimization & Reduction is a low hanging fruit for enterprises to implement which will help reduce the monthly out-go almost immediately helping them to conserve cash and utilize it in other strategic as well as tactical initiatives in these uncertain times.

I have created a 5-Step Cloud Cost Assessment Framework that can guide enterprises to approach Cloud Cost Optimization & Reduction in a more methodical & structured format which will help enterprises avoid costly mistakes and pitfalls while still getting desired results.

5 Steps Assessment FW for Optimizing your Cloud Subscription Costs

1)    Identify:

a.    Identify granular cloud resources and their billing through Azure/AWS/GCP Cost Management services & tools

b.    Specifically, identify and tag high-cost guzzling resources and try to find out the reasons

2)    Classify:

a.    Prepare an asset list of deployed cloud services (and their sizes/tiers) which are Business Critical like Email, etc

b.    Important Production Environments of IT Solutions like SAP ERP, Data Warehouse, etc

c.     Dev-Test Environments, PoC & Demo environments of ongoing IT Solution projects through internal teams or contractor teams

d.    Orphaned or Unused cloud resources which no one knows about and are afraid to delete due to no impact analysis

3)    Analyze:

a.    For Business-Critical & Prod environments analyze actual load & resource utilization against the set size/tier of the cloud service

b.    Analyze if they can be scaled down on size/tier or Instances owing to less average utilization

c.     Analyze if they can be configured for autoscaling by the time of the day/day of the week or by the load on the resources

d.    Analyze if the scaling can be done manually by the team if autoscaling is not possible

e.    Analyze if Dev-Test, PoC, Demo Environments can be shut down during non-working hours and weekends and whether their sizes/tiers are minimal or smallest

f.      For orphaned & unused, analyze if they can be completely removed, or change their size/tier to a smaller unit or move them to free tier or can be archived/stopped/deleted completely

g.    Analyze the technical architectures of the overall IT Solutions running in Cloud and identify possible Rearchitecture & Reengineering candidates which can optimize cloud consumption even further post the re-architecture/re-engineering exercise for the long run

4)    Consolidate:

a.    Different departments, business units, subsidiaries may have started utilizing Cloud services at different time periods and may have subscribed independently thus having multiple subscriptions and billing

b.    As a central entity, enterprises need to consolidate different subscriptions under one common corporate account or subscription for better management, administration and negotiating power with the Cloud Providers

5)    Negotiate:

a.    For Business-Critical & Prod environments negotiate for long term commitment pricing with the Cloud Providers

b.    Negotiate on basic rack rate with Cloud Service Providers or directly with Cloud Companies owing to higher number consolidated resources

With a quick 1-2 weeks of Consulting exercise based on the above 5 Steps Assessment Plan, a detailed consulting report can be developed advising on:

a)    Ways to reduce the cloud costs

b)    Probable monthly savings expected

c)    Detailed implementation plan

For any technology consulting or tech-blog writing engagements, please feel free to drop me a message on Linkedin.

-Pankaj