With so many pluses and minuses in the decision to go to on-premise or into the cloud, many organizations are opting for a mixed approach. Typically, “the challenges for managing the data on the cloud, locally, or in a hybrid environment are not radically different,” said Fritchey. “Getting the right data into the system in an efficient manner, keeping the data, and then getting it back out of the system in an efficient manner remain the fundamental challenges.” What changes are not the problems but the solutions, said Fritchey because different methods are used to deal with issues on-premise or in the cloud. “The fundamentals though are the same,” he said.
However, preparing for a hybrid approach needs extra attention. Organizations are orchestrating data between two different environments, but the data redundancy offered by the free movement of data and applications between cloud and on-premise outweighs the associated difficulties, said Roguine. It can be challenging to split between on-premise and cloud, but that comes down to doing an efficient job of mapping out the entire IT infrastructure. “Once you know which datasets and systems are most mission-critical, then you can decide in which storage site they live,” said Roguine.
The key issue when splitting up data, whether on-premise, cloud, or in multi-cloud scenarios, is “governance, visibility, and control,” said Raheja. Going hybrid ends up increasing the data silos that many companies have tried to solve in many ways, whether using a consolidated location with a data warehouse, a data mart, or data lakes, he said. Spreading the data out presents a different challenge in terms of integrating the data back when needed for running business processes and analyzing the data.
Splitting data between on-premise and cloud environments requires a strong data governance program that “establishes, manages, and communicates data polices, definitions, and standards for effective data usage,” said Marshall. “This ensures that once data is decoupled from its source environments, the rules and details of the data are known and respected by the data users. The point of the data strategy is not to limit data accessibility and access, but to ensure the data represents the best data inside and outside of the cloud.”
THE MULTI-CLOUD APPROACH
Just as vexing as making decisions about splitting data between on-premise and cloud environments is how to split data between multiple cloud providers. “Multi-cloud environments are becoming increasingly popular to not get locked in by a single vendor with respect to pricing and to provide significant business continuity,” said Monte Zweben, CEO of Splice Machine. “For example, an enterprise may have an application in production on AWS powered by a Splice Machine RDBMS and machine learning platform, but replicate to Azure for disaster recovery. This gives business continuity beyond rack, availability zone, and region redundancy for a cloud provider—it enables businesses to continue when a single cloud provider is a target of a major cyberattack.”
While multi-cloud strategies provide greater data portability and higher availability, they “create challenges for companies in data governance, multiple-skill-set requirements, multiple-vendor management, compliance, and security,” said Kumarasamy. In addition, relying on multiple cloud vendors means needing skills for different cloud platforms.
The hidden and not-so-hidden charges that crop up as data is moved between clouds also can add up. “An issue with multi-cloud setups is cost, since public cloud subscriptions often charge for restoring, accessing, or moving data, which can add up quickly,” Roguine warned. Data egress costs are an important consideration in multi-cloud decisions, agreed Manoj Karanth, vice president and global head for data science, engineering, and digital business at Mindtree. “While cloud providers do not charge for data coming into the cloud, they charge for data going out.” The more the data systems talk across the clouds, the more the costs will increase.
Still, cloud providers are, for all intents and purposes, vital partners to many organizations, particularly as their digital transformation strategies evolve. “Many organizations have strategic alignments to specific cloud vendors and try to consolidate all their solutions in a single cloud provider,” said Marshall. “However, some departments will make a business case for different providers, so companies will have a variety or mix of cloud providers. This might happen due to geographic coverage, as not all cloud providers cover the same regions, or exceptions might be made if a specific application is highly valued. For organizations that don’t have strategic alignments in place, they may start with one provider and move to another for different reasons—cost or skills—resulting in solutions in multiple clouds.”
MAKING THE BUSINESS CASE
There are clear benefits that the business may see when data is maintained in multiple clouds or locations. “You have redundancy,” said Fritchey. “Short of a meteor strike taking out half a continent, the differences in locations between one cloud vendor and another means that, even if you’re storing your data within the same general geographic region, you have redundancy. If the first vendor goes offline, there’s a strong likelihood that the second vendor is still available.”
Increased choice is another benefit that percolates up to the business. “Multi-cloud strategies allow organizations to select different cloud services from different cloud providers because some services provided by one cloud vendor are better than others,” said Kumarasamy. “Organizations look at multi-cloud as way to provide high availability services for their customers. In addition, organizations adopt multi-cloud strategies because they don’t want to be stuck with a single cloud vendor and experience vendor lock-in.”
The business case for deploying data or databases across multiple clouds “must map back to the longer-term strategic goals which the company has for digital transformation,” said Janet Liao, product marketing manager at Talend. “If implemented successfully, multi-cloud supports the enterprise by providing agility and flexibility with regard to digital transformation goals such as delivering innovations and operational efficiencies.”
In addition, the business case must also “factor in multiple use cases across departments and consolidating it on an organizational level,” said Liao. “If the business case is built based on a use-case-by-use-case basis, management may not find it worth the time and investment to pursue it.”
Ultimately, every data and cloud arrangement will be as different as the companies that implement them. “While some organizations are looking for standardized, off-the-shelf, subscription-based solutions to meet their cloud-first strategy, other organizations are looking for proven solutions and infrastructure tuned specifically for them that reduce the burden of IT teams,” said Marshall. “Either way, moving to the cloud ensures software and supporting infrastructure remains up-to-date and that data is easily accessible and maintained.”