Cloud workloads see gradual uptake, but miles to go for databases
A picture of steady but not rapid migration to the cloud by Oracle users emerged at Collaborate 18, with database vets saying it's wise to be selective on what to move.
LAS VEGAS -- Like the poet Robert Frost, who had many "miles to go" before he slept, the move to the cloud for Oracle data professionals proceeds, if at a fitful pace.
Issues of integration and project scoping are among many hurdles established organizations still face 10-plus years after cloud computing arose at Amazon, Google and other web companies that built up growing farms of distributed server clusters and made them available for customers to use.
A picture of steady, but not startlingly fast, cloud progress emerged at Collaborate 18, the yearly community gathering of Oracle users put on by the Independent Oracle Users Group (IOUG), Oracle Applications Users Group and Quest International Users Group. Successfully moving databases and applications to the cloud means selectivity in choosing cloud workloads to migrate, according to the database veterans taking part in the event.
The extent to which one database or application is integrated with one or many other databases or apps is an important issue to consider in such selections, according to Kai Yu, a technical staff member and product development leader at IT vendor Dell EMC, speaking as part of an IOUG Collaborate session concerned with successful strategies for what the conference agenda dubbed "the Oracle cloud journey."
"What I see is some applications that are highly connected and some that are relatively isolated," Yu said. "The isolated ones are the ones that go to cloud first." The higher complexity involved with multiple connections can complicate the move to cloud and cause surprises that are not welcome, he elaborated.
Replication in increments
Deciding how much to move in what kind of sequence is an issue that needs to be addressed in moving to a cloud workload. So, scoping out a data function or process that can be well-defined is another way that people are choosing to begin cloud data migration, according to Arjen Visser, founder of Dbvisit Software, a maker of data replication tools with U.S. headquarters in Santa Cruz, Calif.
"It is difficult to get downtime on databases these days, and approaches that move data in increments can be useful," Visser said in an interview. In the incremental scenario, users can "just replicate into the cloud and slowly migrate over to cloud systems over time," he added. Another way to step up to the cloud is to use it as a platform that hedges against natural disasters, ransomware, hacking and other attacks.
"A good place to start the journey is to host your disaster recovery into the cloud," Visser said. "Eventually, the system in the cloud could become the operational system."
There are caveats: Swapping data center workloads for cloud workloads can hold peril. For example, in such replications, "the operating system and the database have to be the same," Visser explained. "That can be tricky to migrate smoothly. It may take a couple of steps."
Two tracks to cloud
For larger companies, it's going to take longer to get to the cloud.
Charles Kimco-founder and president of Viscosity North America
For existing enterprise applications and a legion of databases supported in the well-established large companies that form one of Oracle's biggest customer bases, the shift to cloud has been -- and will continue to be -- gradual.
Quick clouds continue to be the province of smaller companies and newer applications, with larger and more established companies running on a different track, according to Charles Kim, co-founder and president of cloud and database consultancy Viscosity North America in Carrollton, Texas.
"For larger companies, it's going to take longer to get to the cloud. The bigger you are, the longer it is going to take," said Kim, a longtime Oracle database professional.
The gradual uptake comes despite a general edict from many CEOs to move to cloud implementations, he added. Shifts to cloud workloads, it seems, may often continue to be works in progress.
