These days, clouds are everywhere, providing today’s database managers with an impressive range of options to choose from—including public cloud, private cloud, and, for most, somewhere in between in the hybrid realm. There may be multiple variations within a single organization, and these distinct hybrid environments are constantly evolving as well. Ken Rugg, chief product and strategy officer at EnterpriseDB, describes these as “intentional” and “accidental” hybrid environments. Many of the accidental hybrid environments are arising as various groups within organizations subscribe to their own cloud services without a centralized strategy.
Accidental or not, “variety” is the watchword for many hybrid projects. “We have customers who are running database operations in a single cloud, and then others who are running in hybrid fashion—part on-premises and part in the cloud,” said Robin Schumacher, SVP and CPO for DataStax. “A number of our customers are running in multiple clouds, with different databases on different cloud providers. We have some customers involved in inter-cloud deployments, where they have one logical database that spans multiple cloud providers in an active-everywhere fashion.”
Use Cases
Hybrid clouds are appealing because it often isn’t feasible for companies to move 100% of their data operations into the cloud. “We have encountered customers desiring all of the benefits of a cloud solution but who cannot move their databases to the public cloud due to sovereignty laws, industry regulations, corporate policies, or security requirements,” said Maria Colgan, master product manager of Oracle Database. “We often see customers running applications that are subject to sovereignty laws or industry regulations on-premises while they run database environments for their remaining applications in the cloud. Database systems in the cloud often include departmental transaction processing systems or data marts, sandbox environments for data scientists, as well as development environments for new applications.”
While there are a myriad of cloud choices and configurations, there are some commonalities seen with hybrid cloudified data environments. Manoj Nair, chief product officer at HyperGrid, identified some of the most common use cases. Public cloud, he said, is often used for remote disaster recovery sites, as well as associated databases or backup and archiving of database data. Other scenarios include “hybrid cloud database environments leveraged within a transition process by companies in the process of migrating some of their infrastructure farms or application and associated databases.”
Additionally, Nair pointed out, there are hybrid cloud database environments “in which a workload continues to run legacy parts on-premise—a legacy CRM system—while a modern app, such as a mobile storefront for retail, is built on the cloud.”
Paul Ponzeka, CTO at Abacus Group, sees two major types of customers moving to hybrid database environments. “The first is the customer that has moved dev or QA [quality assurance] to the public cloud, but still wants to run their production system on-prem. The cloud gives them some flexibility in running resources when they are needed according to their development lifecycle.” The second type of cloud customer, he said, “is the customer that has begun adopting cloud-friendly database systems, such as web-based apps or cloud-native database applications, coupled with legacy database or fat-client systems. The legacy database setups usually don’t work as well performance-wise in the public cloud, so this gives them the best of both worlds.”
Many companies “leverage cloud as a host for the innovation layer for the core system of records or for certain line-of-business applications, allowing for introduction of new and better capabilities to users quickly,” said Vishal Awasthi, SVP of technology for Serrala. “This enables them to keep up with changes in the business and keep users happy while also giving the IT team the time they need to modernize the core functionality in the ERP system.”
Achieving Balance
Still, not every application is cut out for the cloud. Some are best left on-premise. Colgan pointed to “applications that are subject to sovereignty laws, industry regulations, corporate policies, security requirements, and network latency” as examples of such applications. In addition, any applications for which it is impractical to move databases away from other tightly coupled on-premise IT infrastructures are good candidates to remain in on-premise database systems, she added.
Of course, the decision behind cloud should always be based on the application it is intended to serve. These decisions may depend on the relative age of the application, and in many cases, the best cloud candidates are “greenfield or new applications driving new innovation for their business,” said Nair. “Large historical data repositories will continue to live on-premises, while the intelligence and data modeling outputs from the data are fed into newer cloud applications living in the public cloud.” This means that data may need to be maintained in a hybrid environment. “As a result,” Nair said, “systems like iPaaS [integration platform as a service] approaches are used to best share data across the workloads in hybrid applications.”