Understanding the public cloud billing model before crippling your company

Alan Browning

Tuesday 25 July 2017

Many companies are moving their traditional data centres to the cloud, but public cloud billing methods could potentially bankrupt as many businesses as they enable.

Public cloud is being touted as the next big thing that will drive enterprise business. However, unless you understand the billing metrics used, businesses run the risk of incurring more costs than they can possibly sustain.

Breaking down the costs

Most people are aware that the billing model for the cloud is based on a consumption charge, where most cloud providers have adopted a model of per hour per GB for CPU, RAM and storage. At first glance, $0.01 per GB seems incredibly enticing, and the promise is that you can spin the machines up and down – especially during non-core business hours – and therefore manage costs.

However, the one component that you will be constantly billed for is storage. The methodology here is that if I am using the storage, it can’t be resold to someone else. Hence you will be paying the storage costs, albeit at a ‘cheap rate’, for the rest of your life. There are additional charges that are permanently billed for, such as deploying additional VLANs, which means the costs quickly add up.

But here’s the real kicker: anyone ever heard of egress data? Let me explain.

Cloud computing is often referred to as “the Hotel California approach”. To (mis)quote the famous song: “You can check in any time you want, but you can never leave.” In this context, ingress data (data moving to the cloud) attracts zero tariffs, but egress data is the exact opposite. It means every time I get my data out of the cloud, it attracts a charge.

Let us presume my company of 1000 employees is running at 100 per cent on a public cloud provider. I have very wisely negotiated a deal with my connectivity provider that my egress data is charged at two rand (R 2) per MB. Imagine I create a PowerPoint presentation of 10MB (which is not unusual). Now, five of my colleagues want to review the presentation and therefore must download a copy. All of a sudden, my egress data is charged as follows: R 2 x 10MB x 5 people = R 100. Doesn’t sound too bad, but extrapolate this across 1000 users and it becomes blatantly obvious that this hidden cost has the potential to really stack up – and unless properly managed, it can spiral out of control.

Is this the end of the charges and risks?

Living in a country that has a currency other than American dollars (USD) poses challenges of currency fluctuation, especially if you are billed in USD. Again, another real-world scenario: being based on the southern tip of Africa, our unit of currency is South African rand (ZAR).

In 2012, the ZAR to USD exchange rate was roughly R 8 to US$1. Fast-forward four short years and the exchange rate hit R 14 to US$1. This means our currency depreciated roughly 80 per cent over four years, which means my IT budget would have had to increase by 20 per cent year on year to do exactly the same (not more) compute in 2016 as what I was doing in 2012. Not very plausible. Having worked for a large telecommunications company, as well as a large financial services company, I can confirm that our IT budget decreased by, on average, 8 per cent year on year.

Numerous other threats exist, such as data privacy issues and increased operational costs with running a private cloud and using public cloud services. This creates huge issues in terms of accountability. Cloud makes sense if you are all-in or all-out, but the concept of a hybrid cloud is fraught with danger.

And let’s not forget the elephant in the room: how do I insource my services when I realise the public cloud is not all it’s cracked up to be?

What’s the solution?

Glad you asked. It’s really simple to deploy the correct HCI solution, scale computing, storage and more by using Lenovo products and our partners’ solutions. They can give you all the flexibility – essentially, the benefits of public cloud without the risk.

YOU MIGHT ALSO LIKE