Cloud Platform Economics
With the advent of the Cloud as an industry platform, we now have the concept of Cloud Economics.
Are you financially savvy with your cloud?
Let’s start with the primary principle in Cloud economics: the exchange of CAPEX for OPEX, or in layman’s terms; you trade your capital expenditure (aka. CAPEX, buying the hardware) for variable or operational expense (aka. OPEX, spinning services as needed).
With the Cloud you get ON-DEMAND everything, from large scale computing, networks and databases for almost no sunk cost and everything delivered NOW.
For the tech startup or the company’s IT intrapreneur, this will empower experimentation without the need for intricate planning and logistics,
Cost of Ownership
Before doing that leap to a cloud service provider (CSP), the savvy economist needs to understand TCO (Total Cost of Ownership), which is the benchmark for anything related to Cloud. It is the main metric to allow you to measure the ROI you are seeking with your cloud migration.
You will use the TCO to compare a variety of cloud solutions: AWS, Azure or GCE and even hybrid or private setups.
Your TCO analysis will start with your bare-metal setup and its cost:
- Electronic hardware and all its required support (cables, cooling, etc.).
- Utilities, especially for running that over-provisioned hardware and cooling.
- Labor costs for everything.
- Depreciation and replacements (the average server lifespan is 3 years).
Over/Under-provisioning is one of the headaches you will solve with a cloud migration. The following chart will highlight this.
The cloud is truly adaptable and agile.
Buying your own hardware you will always incur that opportunity cost, being over-provisioning (waste) or under-supplying (loss of clients), while using the Cloud’s resources and setting yourself to scale with them, will result in no opportunity cost and other benefits:
- Match supply and demand via an elastic cost model.
- Match compliance or legal requirements (e.g. AWS is certified in many ISOs)
- Detach from hardware maintenance and datacenter worries.
- Totally remove unplanned outages and control planned ones (unless google or amazon go down in a world ending event)
- Improves your SLAs.
- Generally improves your business agility for time to market, operations and cost of innovation.
Business Agility Enabler
Agility goes hand-in-hand with the Cloud as you iterate and experiment.
Say you wanted to try a document-based database instead of your SQL Database to process and transform your data. If you had a bare metal configuration, this is what you need to do:
- Purchase a server powerful enough to run your no-SQL database and its data processing software. This will take an investment of $4–8k and a couple of months to be shipped to the data center and installed.
- You will test your analysis models and realized you got no value from this change. You have to decommission the server and sell it, spending more time and effort.
If you were agile and on the cloud, this is what it would have involved:
- Turn-on your cloud platform’s no-SQL database setup, for free, in seconds.
- Run the experiment for a couple of months, incurring a running cost of $1k, or nothing at all if you negotiated some credits with your CSP.
- If you didn’t like it — turn off the service in an instant at no additional cost.
Speaking of business agility, we will bring along the support of the Agile Specialist at VelocityOne with our engagements.
The Frugal Cloud
Agility does not only come in the form of quick pivots done at no cost, but also in utilizing exactly what your business requires.
What economists refer to as frugality, we refer to as Right Sizing, or using your resources at their peak with no waste. In the chart above, with right sizing we would have no opportunity costs.
Right-sizing is an exercise done via hard data. In our experience, this is what we look our for:
- Orphaned resources — that provisioned virtual server lying idle in the corner
- Misconfigured resources — Having multiple servers ready to take load, but only one is configured to be used.
- Over-provisioned resources — Having a 80 vCPU server doing a job that only needs 4.
- Incorrect pricing models — Paying $5/GB for a SSD storage with instant retrieval when you can pay $0.05/GB to store items you barely look at.
Cloudy Costs of Migrating
Even though we are in the business of making cloud transformations, we need to be truthful to our customer with regards to unforeseen migration costs:
- Loss of control — The CSP runs your hardware and network, are you comfortable with this?
- Mission critical workflows — Your business lifeline will depend on the whims of the CSP unless you are ready to pay extra, e.g.15K/month for Amazon to support you.
- Cloud pricing — The CSP dictates the pricing and might lock you in. Lock-ins happen when you change your platform to use specific CSP APIs and you cannot move the platform to another CSP without additional work.
- Professional Services — Given a special skillset is required to run things in the cloud, you might incur charges from their teams to do anything if your people aren’t trained.
There are also cost blind spots that need to be accounted for, granted that most of these will be covered by us and therefore kept under control
- Specific skill set needed. All trainable by us.
- Operating on the Bleeding Edge of technology, which will give you few alternatives once you migrate. We always chose not to lock-in to a CSP.
- Reworks to your product to adopt the Cloud. We recommend to keep these to the minimum.
- Scaling minute operational costs (especially compute and network traffic) that can grow big over time if left unobserved. We will help you controlling these costs.
Don’t let cloud bills bring the end of your business
Book a quick consultation with us to build your AWS infrastructure or cut that pesky AWS bill: info@callforatlas.com
This Article by Adam Darmanin is licensed under CC BY-NC-SA 4.0
