An SMB's Life in the Cloud
This month marks a notable milestone in the history of my employer's IT department. We're decommissioning our last on premises server, and completing the migration of our data center to AWS and Azure. This is a process that began nearly a decade ago, when we launched our first virtual machine on Amazon Web Services.
When I first joined CRMT, they hosted marketing databases in a dedicated data center. Salesforce and Eloqua killed that market during the 2009 recession. Yet many of the legacies of that infrastructure remained, whilst being upgraded to more recent versions as they were released over the years. On premises rack servers were considered the only way to safely and reliably host a domain in a Windows environment. Whilst it has long been possible to host file servers and Active Directory on the public cloud the experience is not optimal. Instead, many small businesses have persisted with hosting their own servers for core domain infrastructure, whilst moving as much as they can to the public cloud. Cloud storage and device management technologies simply haven't been mature enough to meet the requirements of existing businesses with a legacy network. This is beginning to change with technologies such as Azure Active Directory or Jumpcloud.
The biggest shift has been the realisation that unless you have the scale to build an enterprise grade data center, public cloud is actually a much cheaper and more secure option for hosting servers. For this reason, Governments and Healthcare have become leading adopters of cloud computing after a belated start. Customers needing bespoke hardware or performance requirements not available on the public cloud can look at the rapidly consolidating co-location market rather than attempting to build their own server room.
Even in 2009 the cost benefits of scale were significant, but the advances in technology driven by public cloud providers have only increased this advantage. It's just that achieving the required scale requires a 6 or 7 figure investment which is only possible for enterprises or businesses looking to make a profit from server infrastructure, by hosting for third parties.
The main thing holding companies back has been the concern that not all workloads are suitable for running in the cloud. This applies particularly to latency sensitive applications, where a few extra milliseconds lag can have business changing consequences. Such applications are particularly common in financial services, where massive sums of money are wagered in stock trades every second.
This has resulted in the edge computing trend, where data centers are broken up and applications moved so they are physically closer to their end users. Data sovereignty concerns in Europe, China and elsewhere have accelerated this trend. This has opened up an opportunity for the likes of Equinix to pitch their data centers to new audiences. However, the biggest impact of the edge computing buzz has been on Amazon, Microsoft and Google, all of whom have substantially accelerated the pace of new data center openings for their public cloud services rather than just expanding their existing locations as in the past.
Yet for many small businesses, talk of edge computing is meaningless. Their applications don't have the scale to take advantage of the benefits. European businesses appreciate the opening of new AWS and Azure data centres in their home countries though. This resolves one data security concern for them, yet for many SMBs the biggest barrier to cloud adoption is trust in the lack of control over data in the relevant services and the reliability of the Internet connection used to access such services. These questions apply to on premises data centres too. However, cloud advocates have yet to adequately address these concerns, until they do many companies will continue to invest in on premises solutions.