“Hybrid cloud” is one of the hottest terms in technology today. All vendors have a hybrid cloud strategy, but like potato salad recipes, no two are alike. Cisco Systems, Dell, Hewlett Packard Enterprise, IBM, Oracle, VMware and others all have different takes on hybrid cloud, and this presents a challenge, especially in multi-cloud/multi-vendor environments, because few standards exist.
Clouds are a set of pooled computing resources that can be provisioned and orchestrated on the fly in an automated manner. Amazon Web Services is a great example of a public cloud service, and your own business may run a private cloud in its datacenter. Similar to virtualization, clouds are an elastic pool where resources can be added and sliced in any way, all from an interface that exposes the capability to the end user (self-servicing vs. IT job tickets).
Originally, the hybrid cloud concept was a single cloud where compute spanned both public and private domains, existing on both sides of the firewall. But from a security and a logistics standpoint, this was hard to implement. Today “hybrid cloud” is becoming “hybrid cloud environment”, a strategy where businesses might be running applications in different environments, with multiple cloud vendors, both public and private. Data and resources are shared across multiple domains (and providers), but each element only lives in one domain.
Normally compute follows data, but there may be instances where data and compute need to live in different domains, whether it is for security, latency or other factors. The interconnection between these disparate elements of the hybrid environment is where businesses struggle, and cloud service providers step in to join it all together. Here are some basic examples of a hybrid environment:
(Source: John Fruehe)
Typically, private cloud might be used for proprietary or differentiated applications, essentially the company’s “secret sauce” applications that give them a competitive advantage or potentially hold very secure/confidential information that must live inside their firewall. Private cloud is typically not as self-serviceable as public, but that is changing. Private clouds also require hardware although externally hosted options, and pure Opex models are now finding favor. Public cloud tends to be used for application development, “bursty” workloads or applications that are non-differentiated (like the typical backoffice operational and billing applications that work the same for everyone). Both public and private have their place and long term coexistence will be fundamental for most businesses, thus the need for standards.
Clouds differ from virtualization and traditional IT infrastructures in the following ways:
|Public Cloud||Private Cloud||Virtualization||Traditional IT|
|Hosted||External||Primarily internal, could be hosted externally|
|Billed||Opex only||Typically Capex & Opex. Some pure Opex models emerging|
|Shared infrastructure||Yes||Only when externally hosted||No|
|Proximity to firewall||Outside||Inside|
Most companies are entering a hybrid cloud environment because they are dealing with multiple clouds, data sources and vendors. Here are some examples of what vendors might refer to as hybrid cloud environments:
- Running a private cloud in your datacenter while also leveraging public cloud services like Amazon Web Services or Microsoft Azure
- A private cloud application that integrates an external data feed like meteorlogical or mapping data from a public source
- Using public cloud analytical tools to analyze your company’s proprietary internal data that sits within your datacenter (or in a hosted private cloud)
- An Internet of Things (IoT) private cloud application in your datacenter using public cloud services as endpoint gateways for collection of telemetry data
- Using “bursting” to push private cloud apps to a provider when traffic explodes
With many vendors approaching hybrid cloud differently, there need to be some standards, common methodology and lexicon to help businesses navigate this area. Multi-cloud and multi-vendor are becoming the preferred strategies for most companies; interconnection, policy adherence and common management will need to be in place for these strategies to succeed.
The Open Networking User Group (ONUG) is actually working on this challenge, creating a Hybrid Cloud Framework that will enable vendors to not only get onto the same page in how they address hybrid clouds, but also solve one of the biggest customer challenges being faced today. At the last ONUG meeting someone posed the question, “Is cloud just the next generation of proprietary lock-in?” A compelling question to be sure, primarily because most believe that they should be able to move an app from one cloud to another—but few (if any) have ever accomplished that. A common framework would go a long way towards helping businesses work with hosters, brokers and cloud technology providers. The working group will be focused on defining the standards around security, contracts, technical architectures and more, helping to put together something similar to the Rosetta Stone of hybrid cloud, enabling businesses and providers to all work on the same page. Just as standardization helped the server business grow rapidly, some standardization in the hybrid cloud space could make it easier for businesses to make the move into the cloud, gaining more efficiency and flexibility. By not being locked in, a business can leverage clouds without worrying about the downstream complications.
If you are going to be in New York on October 24-25 it would be worth your time to attend the ONUG Fall 2016 event to learn more about how hybrid cloud will impact your business.