• Home
  • /
  • Storage
  • /
  • Cloud is not for everyone but when does it make sense?

Cloud is not for everyone but when does it make sense?

Early adopters don’t always see the milk and honey promised via lifting everything to the cloud, with the variable cost of public cloud being the biggest issue.
Ethan Searle
By Ethan Searle, Business development director, LanDynamix.
Johannesburg, 24 Apr 2024
Ethan Searle, business development director, LanDynamix.
Ethan Searle, business development director, LanDynamix.

This is not a straightforward question due to the many issues, specific to running a business in this country, which need to be factored into deciding on the question: do you put everything into the cloud, or not?

Businesses in South Africa operate in a tough environment. Inflation is not going anywhere − it increased to 5.30% in January this year from 5.10% in December last year.

It gives only slight comfort to note this is considerably lower than historic inflation rates in the country; for example, in 1968 when it was recorded at 8.67%, and its all-time high of 20.70% in January 1986.

Then there is the power crisis − I won't even go there. There is also the national brain drain as professionals with crucial skills leave to escape rising crime and a weakening currency and head for greener pastures. These are mounting pressures that contribute to the erosion of business margins and a fragile national infrastructure.

Certain workloads are suited to the cloud, and cloud providers are in the best position to run and maintain that infrastructure.

Research from Nutanix, in 2020, stated that 84% of respondents in South Africa (compared to 73% globally and 71% across EMEA) were either moving public apps back to on-premises environments or planning to do so, generally indicating a trend towards scaling back on the use of public and hybrid clouds.

Not everyone agreed with those stats at that time, but an Amazon Web Services (AWS) report recently noted cloud adoption in SA is poised to generate R3.9 trillion. It further predicted the local market has the potential to unlock R3.9 trillion of additional economic value over the next decade (2023-2033) by accelerating the adoption of cloud computing.

This research − commissioned by AWS − quantifies the relationship between public cloud computing adoption and economic growth, and forecasts cloud could contribute 4.05% of South Africa's cumulative gross domestic product (GDP).

According to the report, cloud adoption in 2021 made a significant impact on the South African economy, noting it contributed 0.18% to regional GDP, generating an economic value of R14.3 billion. More than 78% of this impact can be attributed to the national productivity gains, or so-called "spillover effects" on the economy.

However, cloud is still not for everyone, with early adopters not always seeing all the milk and honey that was promised via lifting everything to the cloud. This might account for the fact that this same report notes that in 2021, only 34% of companies in SA embraced cloud computing − significantly lower than the adoption rates of 49% in Western Europe and North America.

The variable cost of public cloud is quite possibly the biggest issue − nobody can give a fixed cost as usage and load make the end bill somewhat unpredictable.

My opinion is that certain workloads are suited to the cloud, and cloud providers are in the best position to run and maintain that infrastructure.

Office 365 is a great example. Here's one possible scenario: you could build your own Exchange and SharePoint servers and manage them; therefore, when something goes wrong it would be up to you to sort it out. But why recreate the wheel? Microsoft is better positioned than you will ever be to fix problems that might crop up.

However, in the case of moving line of business ERP system to Microsoft Azure, Microsoft cannot add the same kind of value. It will give some cloud processing space on which to run the app, but when it goes wrong, you are on your own with regards to fixing it because you don't have access to the behind the scenes layers of Azure troubleshooting.

The cost of putting that app into the cloud does not make sense to me. You can buy hardware from Dell that has the best service-level agreement and support, providing visibility into all the layers that enhance the ability to resolve issues.

In short, just lifting and shifting to the cloud does not make sense in changing the underlying compute, while leaving the apps that sit on it unchanged, and is an expensive way to go about doing business.

In terms of cost, it's shortsighted to look at the cost of engaging a managed services provider (MSP) in isolation, when what should be weighed up is the value it can add versus the assured losses that will be incurred due to downtime.

Uptime is a decision companies will make based on how much they wish to spend, but again that needs to be looked at in the context of customer expectations and experience − the latter is the name of the game.

If customers want 24-hour uptime from the business − that's what should determine how much it wants to spend to remain competitive. For example, for a CIO at a big bank, the cost of downtime far outweighs that of engaging an MSP.

Businesses need to be able to scale up and scale down computing power, as and when needed. That said, I don't think moving everyone back to on-premises systems is the answer in this country − not with our power situation.

Data centres are geared to run on generators for days at a time, if necessary, and not all businesses can afford that − this is a significant consideration for not bringing everything back to your own data centre.

What I see happening is businesses provisioning their own private cloud in data centres, as that way they get the benefit of being in control of their systems and having the uptime that these data centres offer.

Our own business is similar to this environment. We use office 365 for all the messaging and collaboration tools; we let Microsoft run that and get on with it. Our ERP system plus systems management tools that we want more control over are hosted in our own private cloud.

Lastly, with the ‘joys’ of load-shedding and overworked generator failure, you can still operate by sending people home and running over 5G − that way you get the benefits of cloud but not at the high cost of running everything in Azure.

In my next article, I will expand on the cost of downtime and the common sense behind getting the right professional advice to assist in deciding what to put onto the cloud versus what is best retained on private servers.