Why is buying cloud so cloudy?

Remember how cloud was supposed to be so much easier all round?
19 November 2019

Amazon AWS Summit at MiCo Milano Congressi. Source: Shutterstock

Cloud computing is a means of buying compute and storage resources from a specialist services provider in a form that is simpler, more flexible, more broadly featured, and more comprehensively managed than would be possible if supplied from any single company’s own IT department.

Well, in theory at least.

That initial statement is true at the most theoretical and methodological level, but it is rarely borne out (or indeed played out) in reality.


One of the major ‘bugbear’ points with cloud that seems to have surfaced recently is the purchase options presented with what Amazon Web Services (AWS) calls reserved instances (RI).

When industry disquiet about the way reserved instances are sold leads to Facebook’s Mark Zuckerberg complaining publicly, then the tech community at large does arguably start to question this go-to-market technique and wonder whether it is AWS exercising a little too much corporate muscle.

For the record, reserved instances are sold in three forms: standard RI, convertible RI and scheduled RI.

Standard reserved instances are just as they sound, a standard set block of cloud storage and compute is reserved and paid for in full at the highest possible discount due to the inflexibility trade-off. Scheduled RI is similar but is reserved for a specific time period. Convertible RIs are less generously discounted because the customer does have more options to change the amount of cloud power purchased.

Remember how cloud was supposed to be so much easier all round? When there’s even a way of ‘selling back your flexibility’ to the provider for a discount, then you can see how things aren’t always so cut and dried.

Determine workloads

This aspect of cloud complexity has given rise to a new breed of vendors who specialize in applying predictive cloud usage analytics to determine workload patterns. Companies such as Densify, Alteryx, BMC, and others all operate in this space. Densify specifically targets the challenges that customers face when they are ‘locked in’ to reserved instance contracts.

BMC is equally vocal about this issue. The company’s Muhammad Raza says that a holistic knowledge of the infrastructure and application discovery requires organizations to understand how those components interact and relate with each other.

“Especially since network performance issues can spread across [software code] dependencies that are otherwise hidden from view. With the predictive analytics solutions in place, organizations can collect data from across the network, analyze multiple data sources and understand how one infrastructure system can impact the other,” wrote BMC’s Raza in a 2019 company blog.

Backing up a little on backup

Cloud backup services are equally difficult to grasp, comprehend and manage, apparently.

A 2019 study carried out by disaster recovery managed services specialist 4sl claims that more than three-quarters of enterprises see handing over responsibility for backup as a benefit of adopting cloud services, although only 30 percent know their cloud service providers’ backup and recovery processes in detail.

If there is any truth in the suggestions being made here, we can say with some certainty that the need for support with backups is clear. There is a clear compliance need to ensure backup data is kept because the vast majority of enterprises have to retain backups for a specific length of time to meet regulatory obligations.

“The desire to pass on responsibility for backup to service providers is understandable – backup environments are becoming extremely complex and the peace of mind that a responsible partner is managing backup can be invaluable,” said Barnaby Mote, CEO, and founder of 4sl.

Mote insists that enterprises need to understand that ‘in the main’ [for the majority of cloud contracts taken out], the standard level of backup provided for infrastructure or software as a service won’t meet their needs. More than likely they’ll need to invest in the expertise or services to ensure their data in the cloud is protected and retained for long enough.

Validation & illustration

To validate the above technology discussion, we can confirm that of the two most commonly used cloud services, Microsoft Office 365 Exchange Online and Google Cloud, backup provision for data are no longer generally recoverable after the standard 14 days agreement period.

It’s important (but also additionally illustrative) to note that some instances of Microsoft’s cloud services come with longer data backup recovery periods of 90 days or more. And for the record, Salesforce offers 90 days, Slack offers none and Box offers 30 days.

This need not be an issue – enterprises who have made provisions for backup beyond their cloud providers’ standard offerings will not be at risk. But let’s also note that in the real world, most enterprises of any size will use more than one cloud service provider.

So, have you put the right data in the right place with the right cloud for the right backup window needed?

That’s reserved instances, workload analytics, and backup complexities then. We haven’t mentioned storage contract complexities, latency parameters for fast-paced real-time streaming application services, input/output (upload/download) intricacies and exactly how comprehensively a cloud provider will ‘wipe’ customer data if they need to delete their cloud instances.

Going forward into 2020 and beyond, we can expect the cloud industry to tell us that more and more Artificial Intelligence (AI) will be tabled to solve these issues. Cloud might become simpler one day, but don’t hold your breath just yet.