Enterprises need to ensure their operations can continue no matter what. For decades, organisations have used redundancy to ensure they don’t have to rely on one single provider. This principle continues to hold for the cloud.
Here, enterprises can use multicloud storage to get the best of all worlds, and to exploit the opportunities that come from multiple cloud setups.
As the name suggests, multicloud storage is a way for companies to use multiple clouds to store their data. While that definition is simple to grasp, there is a hidden complexity within it as enterprises have to deploy processes to distribute and manage the data across numerous clouds.
The clouds in question can be all public, all private, or, more likely, a mixture of both.
An enterprise may use different cloud providers for different areas of their business, such as infrastructure, software or data, for example. Users can store data across separate providers while having the benefits of each one separately, as well as the benefits of having everything connected.
As far as the end user is concerned, it shouldn’t matter where the data is stored. They do not have to know if it is a local file, an Amazon S3 object or an Azure Blob. Neither should they need to be aware of how processes such as authentication, single sign-on, policy enforcement, and search are conducted.
The same goes for cloud-based applications. These should be able to access data regardless of whether the data is stored in the same or another public cloud or on-premise. The same argument can be made for on-premise applications.
Undoubtedly, Amazon Web Services (AWS) is the leader in cloud infrastructure services with an enormous 47% market share in 2018, according to Gartner. But Microsoft and Alibaba have been progressively eating away at this lead.
Microsoft Azure has grown its share as risk-averse late adopters migrate IT systems to the cloud. Many traditional hardware suppliers also have offerings, such as HPE and Dell EMC.
Benefits and pitfalls of multicloud
Multicloud storage has diverse benefits for businesses. Enterprises can pick and choose best-of-breed cloud services across public cloud providers, and allow data mobility to remove anxiety over supplier lock-in.
It also can mitigate supplier risk. If one cloud provider suffers a region-wide outage, then another cloud provider may be still operational. This allows the customer’s business to continue, albeit at reduced rate while the first provider is down.
But there are also a number of common difficulties that enterprises can experience when dealing with multicloud storage. The first one is to create a plan by accident rather than by design.
If you don’t have a multicloud storage strategy, things will get messy quickly as individual decision-makers will choose cloud storage based on their own preferences rather than thinking about the requirements of the whole enterprise. This can mean the IT team not knowing where data is, who owns it or having any control over access controls and encryption policies.
Just implementing multiple clouds by accident can lead to needlessly redundant storage management operations and poor data management where data is not properly backed up or protected.
Speaking of management, enterprises should use common data management tools across multiple cloud storage services instead of ones that only work with a particular service. They should also understand each application’s performance, because failure to do so could result in spending more money on unnecessary storage.
Multicloud storage security failures can also cost an enterprise in terms of finances and reputation. There needs to be a proper security strategy with policies and controls so that enterprises know what data exists, where, when and with what access controls.
Storage use cases
Organisations can use multiple clouds to improve their storage infrastructures. There are a number of ways enterprises can do this:
Backup and archiving: A key use of multicloud storage is for backup and archiving. This approach makes such a use case cheaper, more reliable and simpler. Data is replicated offsite (making it cheaper than tape storage) while retention periods and recovery points are improved.
Resiliency: Having data storage in many locations can go a long way to mitigate the risks involved in storing data. An outage in one cloud storage service should not affect data stored elsewhere, for example.
Compliance: More stringent data privacy rules now mean data has to be stored in a certain geography to meet regulatory requirements. This means most data can be stored in one country with one cloud storage service provider, but a subset of that data may need to be kept on another cloud in another country to comply with regulations in the region.
Different storage approaches
Multicloud storage poses a number of challenges, such as cost management and data migration. That means there are several different approaches to ensuring that the enterprise has the right strategy in place.
According to Brian Wood, director of cloud marketing at Teradata, separating work across environments enables each to play to its strengths and cost-effectiveness. This applies within an environment as well as between environments.
“For instance, low-cost object stores such as Amazon S3, Azure Blob, and Google Cloud Storage are ideal domains for storing low or unknown-value data. On the other hand, higher performance – and more expensive – block storage such as Amazon EBS and Azure Premium storage is best for frequently queried, and thus higher value data,” he says.
Rich Weber, chief product officer at Panzura, says that a new key best practice he sees in enterprises is to put a filer into their physical work locations, and that way, they can have access to the same data that’s there for the compute cloud.
“When people start to move applications and compute workflows into the compute cloud, that compute cloud, regardless of the provider (such as AWS), is no different than a hybrid on-premise site,” he says.
Every time an enterprise reads the data between clouds, it’s effectively the same as reading that data over the internet and that means they encounter egress charges and, suddenly, their network charges can exceed their actual storage charges if they’re accessing that data frequently.
“What the filer does is caches everything locally, reducing those charges and lowering latency. On average we’re seeing that 95% of the time, businesses can forego the need to do a remote cloud read, which saves them money on those egress charges,” says Weber.