Lesson 1: You’ll pay for cheap infrastructure in the end

In the first of my ‘lessons from a decade in hosting’, I want to talk about one of the most recurring conversations I’ve had with customers over the years. And that’s the true cost of cheap infrastructure. 

Cheap infrastructure is a defining example of the old adage ‘you get what you pay for’. But not just from a financial perspective. Go cheap and the impact can go beyond monetary value to the reputation of a brand. 

What’s the biggest fear companies face when it comes to infrastructure? That their infrastructure goes down, causing at best disruption to customers and at worst loss of data, transactions, money, and confidence in that brand. 

Are you likely to have more downtime with cheaper infrastructure? I would like to think the answer to that question is an obvious one but it’s surprising how few customers consider it when purchasing. 

Which is why I believe it warrants talking about. Starting with what hosting providers commit to when it comes to downtime. 

There is no SLA for ‘planned' downtime

In our industry, every single contract I’ve seen has service level agreements (SLAs) that infrastructure providers must commit to. And every single one of those SLAs has a maintenance clause meaning that maintenance and its impact is excluded from metrics and therefore any penalties. 

It’s important the clause is there because infrastructure maintenance is non-negotiable. Regular inspections need to be carried out to ensure that everything is working as it should and to optimum capacity. There are also occasions when maintenance must be undertaken to fix a problem. 

Generally, these instances shouldn’t impact the customer. But, of course, we’re working with technology and it’s not always predictable, so there are going to be downtimes when customers do feel the effects. What’s key is that these times are few and far between. 

Interestingly, based on my experience of conversations with customers, it’s very rare that we’re asked about the number of maintenance-driven downtime situations we have and the results of that downtime. No one seems to consider the long-term cost of maintenance when purchasing infrastructure. And yet, if these instances occur numerous times a year, those costs can start to make initially cheap infrastructure very expensive. 

The odds are stacked against you if redundancy isn’t built in 

As well as asking hosting providers “how many maintenance-driven downtime instances did you have this year?”, there are also hardware specifications that can help customers identify infrastructure that is less likely to suffer maintenance-driven downtime. Specifically, around redundancy.

Take the network stack as an example, the set-up of which can be vital to redundancy in your infrastructure. If you only have one path from your server out to the internet and that path is disrupted at any point, your server will go offline. By specifying multiple ISPs, core routers, switches, and network ports on the server, you are building in as much redundancy as possible right at the outset.

It is impossible to provide that same level of redundancy at a lower price simply because it requires more hardware. But it can stop something as stupid as someone accidentally pulling out the wrong cable or a firmware issue in a switch from causing an outage. So, if you opt for cheaper infrastructure, you are constantly rolling the dice against how significantly you’ll be impacted when downtime inevitably happens. 

By spending a little more, you are placing a much safer bet on the availability of your infrastructure and therefore your investment. 

What’s your experience? Let me know in the comments. And keep an eye out for more lessons to come.