Enterprises are creating and retaining increasing amounts of data, either because of regulatory requirements or because they are hoping to derive future value. Many organizations decrease their storage-related costs by adopting storage as a service, though this approach could likely use fine-tuning.
The common theme with as-a-service offerings is how organizations acquire the technology. The customer doesn’t own the technology used to deliver these services but is charged for its use. In the case of storage as a service (STaaS), the charge is typically per gigabyte or terabyte per month, based on a tier of performance.
Storage-as-a-service options
The three types of STaaS are public cloud, private cloud and colocation, defined as follows:
Public cloud
The public cloud offers STaaS as a shared infrastructure. Public cloud STaaS has been available since AWS introduced S3 in 2006. The object store charges per gigabyte of data stored, with variations in price for reduced availability and performance. Public cloud service providers also have block and file offerings, although block storage is typically restricted to access via cloud compute instances.
As a model of storage service delivery, public cloud storage can deliver limitless capacity, can scale up and down and requires little or no commitment in terms of capacity or term of use.
Private cloud
Private storage-as-a-service offerings are more complex to deliver. Where public cloud storage is shared infrastructure, private STaaS is dedicated to a single customer. This represents a challenge for the service provider, which must ensure that the required capacity exists, but without driving up its own costs in the process.
On-premises STaaS vendors typically require a minimum capacity commitment, such as 100 TB, and a minimum term, which is usually one to three years. However, some vendors will allow lower capacities with terms as short as one month.
Colocation
A third option is to use a colocation service, where the vendor installs the infrastructure close to a public cloud provider or a local network point of presence. The customer consumes storage resources over the network, with the benefit of both low latency and access to shared resources.
This approach delivers a service closely aligned to the private cloud model, with the option to use either dedicated or shared equipment and access platform-specific features like snapshots and data replication. The customer is expected to cover networking costs to access the storage.
Service-level differences
All STaaS offerings come with service-level objectives (SLOs) as part of a contractual service-level agreement (SLA). In the public cloud, typical SLOs are lower than the enterprise, where five or six 9’s of availability is the accepted norm. Missing an SLO generally results in the vendor offering service credits, rather than any financial payment. Businesses might have to build more resilience into their application designs to deliver higher levels of availability when using cloud storage.
Customers using STaaS offerings delivered in the public cloud can expect to see new services implemented as a continual process, based on the service provider’s timeline. The customer has little or no input into planned outages or upgrades.
Service management can be more complex with on-premises STaaS. The customer sometimes has more influence over timing for software patching and upgrades because the hardware is exclusive. However, this might not always be the case when the storage is delivered as a fully managed service.
When negotiating the service contract, organizations should carefully consider the turnaround time for replacing failed components or resolving problems. Turnaround times might not be the same as those with leased or purchased storage.
Data protection is on you
STaaS isn’t typically offered with data protection. In the case of public cloud, the cloud service provider will restore a failed service to the point of outage.
Data protection sits squarely with the customer in all STaaS instances and should be considered an additional cost.
However, there have been instances where cloud service providers have failed to recover data after a hardware failure. Data protection sits squarely with the customer in all STaaS instances and should be considered an additional cost.
Upgrades vary with service type
Enterprises typically refresh their storage hardware every three to five years. These storage refreshes tend to be driven by increasing maintenance costs as the hardware ages. However, storage refreshes can be driven by any number of other factors, such as a need for increased capacity, higher throughput, less power and cooling, and less rack space.
With public cloud and colocated services, the technology refresh is obfuscated from the customer. Over time, providers refresh their services according to their own schedule, offering newer, faster and sometimes cheaper services. Customers can choose to take advantage and migrate to the new services — or not.
For fully managed STaaS delivered on premises, technology refreshes occur as part of a perpetual or evergreen deployment model. The service provider will maintain the storage for as long as the service contract exists. If a technology refresh is required, then the provider will perform the refresh in a way that adheres to the organization’s SLA. Some vendors have products that can deliver in-place non-disruptive upgrades, whereas others might need to replace hardware through a migration project.
STaaS on the balance
Storage as a service can be a great approach for managing costs, as it avoids upfront capital outlays and aligns costs to consumption. However, the vendor retains control and ownership of the equipment. This might be challenging for enterprises that are used to having full platform access and control.
Those who use STaaS will also need to be aware of hidden costs. Data egress fees, for example, can be substantial and are often incurred when moving data out of the cloud or between clouds. However, in some cases, even running a backup can trigger data egress fees.
In some cases, using STaaS might also increase storage latency, particularly in situations where the storage is geographically separated from the workload that uses it. This can be problematic for any workload that requires real-time data access.
Those who are subject to regulatory compliance requirements will also need to be aware of how using STaaS might impact their compliance efforts. Most of the major STaaS providers are sensitive to compliance mandates. However, data sovereignty can sometimes be an issue when an organization needs to guarantee that its data does not leave a particular geographic area.
Brien Posey is a former 22-time Microsoft MVP and a commercial astronaut candidate. In his more than 30 years in IT, he has served as a lead network engineer for the U.S. Department of Defense and a network administrator for some of the largest insurance companies in America.
Chris Evans has worked in the IT industry for more than 25 years. After receiving a B.Sc. in computational science and mathematics from the University of Leeds, England, his early IT career started in the mainframe arena and continued to storage and systems programming paths.
