When it comes to data storage, public cloud is now firmly mainstream. According to a recent survey, close to a third of organisations plan to deploy cloud storage. CIOs and IT directors listed cloud as their highest investment priority after cyber security and risk management.
Cloud storage has long been an attractive option for organisations that prefer not to own or run their own technology. And the growth of hyperscale public cloud providers and SaaS vendors make it easier for customers – especially fast-growing businesses and smaller enterprises – to operate with little or no on-premises IT.
Of course, to move compute and storage to cloud platforms comes with its own challenges and risks. A lift-and-shift approach to cloud deployment means missing out on many cloud benefits. But, as technology has matured, there are more ways businesses can benefit by moving storage to the cloud.Performance
When vendors first started to offer cloud storage, performance was not the most important criteria. Instead, the focus was on flexible storage capacity. This led to an early focus on applications such as backup and archiving, where performance is not critical.
However, cloud providers now offer tiers of performance, albeit at a cost. Vendors provide flash-based options for workloads that demand performance as well as HDDs for where capacity and cost matter more. Cloud providers also offer high-performance computing (HPC) options in the cloud.
Examples include Amazon FSx for Lustre, and Google’s HPC-focused Filestore (based on its 2019 acquisition of Elastifile). Microsoft, meanwhile, markets Azure premium storage for HPC environments.
A second factor in favour of the cloud is that as more data is processed, it makes sense to keep that data close to the compute layer. Even if on-premises storage is faster, the time taken to move large volumes of data to and from the cloud outweighs much of the benefit.
As GigaOm analyst Jon Collins notes, the main reason not to store data in the cloud is where latency needs to be as low as possible, such as in R&D and manufacturing. “It’s applications such as a wind tunnel, where you need the results right next to where you’re doing it,” he says.Scalability and agility
Being able to scale up and down remains one of the biggest draws for cloud storage – customers pay for what they use, when they use it. In the storage context, this is useful to match capacity to seasonal workloads, to bring capacity online for analytics or research purposes, and increasingly, machine learning and AI.
This only works if customers are disciplined enough to turn off storage they don’t use, or to move data to the right storage tier.
But there is another aspect to agility and that is being able to respond quickly to business needs. To spin up servers and add storage in the cloud can take seconds, minutes or hours. Fitting out a datacentre with storage arrays can take months.
Use of the cloud allows customers to experiment with new applications, products and services. As GigaOm’s Collins points out, this is especially relevant with AI. “We should experiment with AI right now,” he says. “You can’t say, ‘We will experiment, but it will take 60 days before we can switch anything on’.”
To move technology to the cloud also allows firms to focus on their business rather than managing storage.Cost
Arguments in favour of cost savings through cloud storage are more nuanced than those around agility.
Early on, too many CIOs believed the cloud would automatically cut costs. It didn’t, and left organisations with higher than expected bills.
Often, this is down to how cloud storage is used. To keep archived data in high performance storage costs money, so does retaining data “just in case”. Cloud cost optimisation, or FinOps, is still a fairly young discipline.
But with solid data management and the right storage tiers, cloud can be economical. This is all the more so when other overheads, from building maintenance and security to disaster recovery, are taken into account. IT management
IT management is another area where cloud has potential to ease CIO workload.
Total cost of ownership (TCO) can be lower, and cloud storage should offer a more modern management layer that incorporates smart technology and AI to manage issues such as tiering, backups and ransomware detection.
Compared to significant legacy on-premise estates, cloud should be easier to manage. At worst, a firm will have a handful of cloud providers, against perhaps dozens of hardware, software and tool vendors on-premise.
That said, this is an area that is still a work in progress. Hyperscaler management tools are good for their own storage stacks, but we are still some way from a single pane of glass to run all cloud instances. Firms also need to understand distributed architectures and environments such as containers.
The potential is there, however, and IT teams are better off investing in newer developments in the cloud than spending resources on legacy technology. Security and resilience
Security and performance are two areas where cloud storage has really improved over the past few years.
Early cloud systems were far from secure. But now, data stored in the cloud is at least as secure as data on-premise, but only if cloud systems are properly configured. The large cloud providers have invested significant sums, and arguably have better security than many customers, with the possible exception of banks, government and defence.
Customers can leverage that investment to secure their own data, usually at a lower cost than building equivalent security measures in-house, and cloud providers will keep up with new threats too.
Alongside security is resilience. Cloud providers have every incentive to avoid outages. For their customers, failover facilities and backups should be cheaper and easier to manage in the cloud.
Even if organisations prefer to keep their primary data storage on-site, to use the cloud as a backup location makes increasing sense. Done well, it should be cheaper and more secure than most alternatives.
