In his recent book, Migrating to the Cloud: Oracle Client/Server Modernization, co-author Tom Laszewski makes the case for moving your legacy Oracle infrastructure to the cloud. Laszewski, director of the Oracle Platform Migrations Group, talked in this Q&A about how some Oracle pros are running in the cloud without knowing it, why a server refresh is the perfect opportunity for migrating to Oracle in the cloud, and some specific challenges for large database migrations.
A survey of our readers last year found that about two-thirds had no plans to deploy Oracle in the cloud. Why do you think that is, and how can it change?
Tom Laszewski: Cloud means many things to IT professionals and there are many IT professionals that equate hypervisor-based virtualization with cloud computing. IT professionals may consider Software as a Service (SaaS) or Database as a Service (DaaS) in a public cloud environment such as Salesforce.com or Amazon DaaS as what it means for deploy workloads to the cloud. What they don't consider is that in many cases these SaaS or DaaS deployments are running on an Oracle Database or Oracle Fusion Middleware. If a customer is running Salesforce.com, they are running Oracle workloads in the cloud today and do not even realize it. DaaS and SaaS are the most visible aspects of cloud workloads, and Oracle currently does not offer SaaS or DaaS in a public cloud deployment. Therefore, Oracle is the backbone of the cloud but currently not the 'face' of the cloud.
Secondly, IT professionals oftentimes confuse placing an application on a hypervisor like VMware vSphere as a way to move workloads to the cloud. What IT professionals don't consider is moving Oracle Database or Fusion Middleware to a shared computing environment accessible through the Internet as a cloud environment. So, they may be running Oracle Platform as a Service (PaaS) or Infrastructure as a Service (IaaS) in a private cloud setting but don't consider this is running Oracle workloads in the cloud. In some cases, it comes down to semantics and looking beyond just public cloud DaaS and SaaS, or taking the time to understand the technology behind the DaaS or SaaS they are running.
What is the difference between deploying Oracle applications in the cloud and deploying Oracle database in the cloud? Are Oracle pros more willing to do one cloud migration over the other? Why?
Laszewski: Deploying Software as a Service (SaaS) or Database as a Service(DaaS) is more a matter of what Oracle pros we are talking about then which one pros are more willing to adopt. Oracle Database pros will feel much more comfortable with DaaS because they are database-focused. Oracle Fusion Application pros will most likely be more willing to entertain SaaS as they are proficient and comfortable with Oracle Applications. Oracle Database pros will also deploy DaaS for unit, system, and customer acceptance testing so they will be more likely become have more comfort and have expertise with DaaS. This will lead to them lean toward DaaS for production as well.
Let's say I'm an Oracle pro and I'm due for a normal server refresh. Convince me that I should consider putting my Oracle workloads in the cloud instead.
Laszewski: The normal server refresh cycle is actual one of the best times to put Oracle workloads in the cloud. The Oracle Exadata and Exalogic engineered solutions contain all the server, and software, to deploy a public, private or hybrid Oracle cloud solution.
When undertaking cloud migrations, which important steps and safeguards are IT pros most likely to forget?
Laszewski: There are two most important steps that IT pros forget when moving to the cloud. First is the impact of running in a shared-resource environment and how to chargeback users. Compute resources, including even power and floor space, need to be metered and charged to the appropriate consumers. Since the business community is not accustomed to chargeback, the first step can be to showback (show the calculated charges) to the business users and then implement a full chargeback.
Secondly, testing and production rollout are underestimated. Running all the unit test cases, testing the end to end system, and running all customer acceptance use cases is not something that can be completed in a couple weeks or even months. The production rollout is sometimes just looked at from the perspective of the 'week end' switch over. Production rollout involves several test cycles and running the old and new cloud environments in parallel, both of which can consume months of effort.
There are also some overlooked safeguards. Integration in the cloud involves integrating with IT data center applications or databases that are not moved to the cloud. This means opening up ports to the Internet and moving data across public networks. Also, applications and databases that were once dedicated to a department or customer are now shared across clients. Depending upon their area of expertise, an IT professional may consider multi-tenancy at the network, server, database or application tiers. However, an architect or IT management must look at support for multi-tenancy at all layers of the IT infrastructure.
What are some challenges specific to large database migrations?
Laszewski: The two biggest issues are loading/reloading data and the 24x7 business environment. Loading and reloading data during the multiple testing cycles takes a considerable amount of time and compute resources. Meanwhile, the 24x7 business environment means there can be no shutdown of the application to migrate the data and test the migration of the data as systems cannot be down at all.