December 10, 2014 | Written by: Payal Chakravarty
Share this post:
Hybrid cloud—the combination of using public and private cloud to host your applications and infrastructure—is becoming the de facto standard for cloud adoption rather than a transitory architecture. According to several industry surveys, 70 percent of organizations are using or evaluating the hybrid cloud today. There are two distinct directions in which the move to hybrid cloud is occurring: private to public cloud and public to private cloud.
Private to public cloud
Enterprises with big IT investments in their on-premises data center are gradually moving certain workloads to the public cloud to gain efficiency, scalability and agility. Those workloads that have sporadic demands are most commonly shifted to the public cloud. Bound by privacy and security restrictions, these enterprises tend to keep their data on-premises in their private cloud.
Public to private cloud
Companies that were born in the cloud and utilized public cloud to scale without spending on an IT organization are realizing that, over time, the public cloud costs add up. It’s like your daily coffee bill. You think you are only spending four dollars every day, but when you look at the cumulative bill at the end of the year, you will be shocked to find out how much you are spending on coffee. You might as well have bought a coffee maker.
(Related: Want to build, manage and secure a hybrid cloud solution? Learn more at IBM.com/cloud/expertise.)
Another aspect of the move from public to private is that once user demand attains a steady state, the need for auto scaling is not significant. You can save significantly by running a private OpenStack in your own data center instead of running full scale production on a public cloud. These realizations have dawned upon several companies that have started to locate some of their workloads in-house to improve cost efficiency and gain more control over their data and resources.
The movement of workloads from private to public and vice versa is necessitating the need for a stable hybrid architecture.
With this movement, companies must investigate how to span applications across such a hybrid environment. An application typically runs on a stack that comprises:
• Front-end code, which may have dependencies on third-party services
• Middleware such as application servers, web servers, message queues and so forth
• Data sources such as SQL or NoSQL databases
• The operating system
The front end, which needs to auto scale to tune to user demand, is typically the best candidate for being hosted on the public cloud. The middleware and database can reside anywhere. However, the trend indicates databases are the most popular candidates for keeping in house.
To manage the lifecycle of a hybrid application, you need to be able to deploy, manage and monitor across the hybrid cloud. Several tools are evolving to handle this shift.
Tools like IBM Cloud Orchestrator let you define hybrid patterns to solve the deployment problem. This means you have a set of application resources and configurations that spans across your public cloud like IBM SoftLayer and your private cloud based on VMware. You can capture that mapping and configuration as a pattern and deploy that pattern in five minutes across both environments.
Additional tools like IBM Application Performance Management (APM) now let you manage and monitor your hybrid applications with stack components that reside anywhere. If your application’s front-end code is hosted on a public cloud like Amazon or IBM SoftLayer and its back end is a Tomcat server and MySQL database that resides in your own data center, IBM APM tools will discover the two and let you visualize them together in the same pane of glass so that you can see your application’s performance end to end.
Are you deploying and managing apps on a hybrid cloud today? I’d love hear about your experiences and challenges. Leave a comment below or get in touch with me on Twitter @its_pi to start the conversation; let’s talk APM, cloud and DevOps!