While many organizations say they have a hybrid cloud, in fact, they just have separate public and private clouds....
By submitting your personal information, you agree that TechTarget and its partners may contact you regarding relevant content, products and special offers.
This setup doesn't take advantage of the true values of cloud computing: orchestration, self-service automation, and the ability to scale up and down on demand by shifting workloads between public and private.
In the chaos of the new computing era, which is partly due to hype and confusion, the true hybrid cloud definition has lost its meaning. Both public and private clouds are built on a clustered COTS-based architecture that follows a standard method of automated orchestration and the ability to scale workloads by adding or subtracting VM instances as needed. However, some so-called "private clouds" consist of legacy gear -- such as mainframes -- that allow app flexibility, but don't meet the COTS definition. This becomes a significant issue when we look at the hybrid cloud.
What distinguishes a true hybrid cloud from just a pairing of public and private cloud is automated orchestration across the boundary between the cloud types. However, this is virtually impossible if the private side is legacy gear, meaning that a hybrid must all be COTS-based.
The three things that make a true hybrid cloud
A hybrid cloud should meet three key criteria:
- The hybrid should have common orchestration across the public-private boundary to permit the seamless expansion of instances into the public space for load scaling.
- The orchestration system should assist the movement of data between clouds.
- Orchestration should allow customers to switch cloud providers to get better pricing or service.
The first criterion is just being realized today. An extensive set of offerings exist for common orchestration -- but there is a rapid evolution in progress and a few remaining wrinkles to iron out.
Major systems vendors such as IBM, Dell and HP meet the common orchestration requirement. In addition, major cloud providers Microsoft Azure and Rackspace, as well as VMware and Red Hat, meet this criterion. Tools from startups such as Egenera, Eucalyptus and more are also up to the task.
When we look at the second two criteria, however, we begin to see major deviations in the market from a true hybrid cloud. VMware and Azure, for example, bridge data to their own public clouds. While this approach provides a more seamless operation, users relinquish their ability to shop for the best rates and lose the opportunity for savings with rates dropping almost daily.
In addition, having a data management strategy is critical for a sound hybrid cloud implementation. However, data management is still a weakness that needs more emphasis. Slow data rates on U.S. WAN links limit the ability to move a lot of data around. For example, an elastic app that can expand to a public cloud needs access to 3 terabytes of data to run, but moving that 3 TB on demand would be a time and cost killer. We need a method to have the data available in the public cloud, and a method for syncing to the instance in-house.
So, which vendor hits all of your hybrid cloud must-haves? Dell is saying the right things with a current product, and Red Hat is road-mapping a vendor-independent service. The remaining vendors are restrictive on private cloud support, but could be closing in on similar broader connectivity. VMware and Azure may be half-hearted in expanding connectivity due to the desire to sell their own services, but VMware would do well to open things up. The idea of vendor lock-in is becoming a thing of the past, as parent EMC is seeing with arrays.
Use cases for the hybrid cloud
From archiving to load-boosting, there are many use cases for hybrid clouds and many views on which data must remain in-house. In spite of complicating the thought process for implementing a hybrid cloud, the differing views emphasize the need for flexible tools and a good strategy.
Going beyond plain hybrid clouds, we enter the world of software as a service (SaaS) mash-ups. Pay-for-play models are changing the face of infrastructure, and the advent of big data services for rent in the cloud is leading to more of an end-to-end service residing in various clouds outside of the user's home cloud. Handling mash-ups adds more emphasis to the data management issues, as well as creating pressure for integration tools to bridge apps together quickly. Expect a rapid evolution and a lot of competitive action in hybrid and mash-up management.
VCloud Air sends VMware in right cloud direction
Management tactics for the hybrid cloud
Does hybrid cloud work for your business?