By Jason Dover
Editor's Note: Occasionally, Enterprise Networking Planet is proud to run guest posts from authors in the field. Today, Jason Dover of KEMP Technologies discusses the advantages and challenges of a hybrid cloud.
"Even though you might not realize it, over 95 percent of you are already consumers of cloud computing services." I vividly recall this opening statement by a speaker at a tech conference several years ago, just after he asked everyone to answer by a show of hands whether they were Yahoo and Gmail users. We're already at a point in the lifecycle of cloud computing where it is beyond the phase of nebulous marketing hype.
The formation of the cloud and the need for the hybrid cloud
Although the mid-2000s marked the beginning of cloud computing as a household name, the concepts were born more than five decades ago. In the 1950s, mainframe computing laid the groundwork of pooled resources in a cloud-like infrastructure shared by dispersed users. Technologists of the 1960s and 1970s introduced the world to a vision of an interconnected globe with access to easily scalable programs, resources and data, regardless of location and without the bounds of a rigid system infrastructure.
Fast forward to 2002 and then again to 2006, when Amazon delivered a resurgence of this notion with the development of Amazon Web Services (AWS) and then Elastic Compute Cloud (EC2). This made possible the delivery of cloud-based storage and compute so companies can rapidly provision services without large capital expenditures or the limitations of in-house infrastructre. Since then, Infrastructure as a Service (Iaas), Platform as a Service (PaaS) and Software as a Service (SaaS) frameworks have multiplied by an order of magnitude. IT decision-makers now have endless options for leveraging public cloud service offerings to augment their overall IT delivery strategy.
Despite the advancements and benefits in public cloud computing, however, governance implications, economics, concerns over reliability and out-of-the-box support for custom business-critical applications have staved off adoption of a public cloud-only model by an overwhelming majority of organizations.
These limitations have become a main driving force for a hybrid approach to cloud computing. Hybrid cloud is often oversimplified as merely being an environment that leverages public cloud infrastructure for some applications and on-premises infrastructure for others. While this definition is true in the strictest of sense, it fails to convey the depth of the expected outcome of building a hybrid cloud infrastructure in the first place: integration of heterogeneous services both in front of and behind the corporate firewall with such symmetry that each entity behaves as part of a bigger whole.
What to expect when you're expecting a hybrid cloud
Actual execution isn't easy. Here's why. Successful hybrid cloud implementation assumes a well-architected private cloud as opposed to simply a well-built traditional IT infrastructure. Adoption of hybrid cloud starts with the transition from a traditional on-premises environment to one that includes concepts and supporting technologies to enable functionality normally associated with public cloud: self-provisioning for application owners, dynamic resource scaling, a chargeback model for lines of business, orchestration for automating repeatable tasks and a high-visibility management platform to monitor how and where services get deployed.
It's the familiarity with the very nature of the public cloud model that has fueled the business and technical requirements in the enterprise for what is essentially an IT as a Service (ITaaS) framework that allows for agile self-service, provisioning and consumption monitoring while simplifying the load on application owners. Because many legacy data center environments were not built with these principles in mind, transitioning can be a challenge.
Hybrid cloud also opens the possibility for workload overflow processing or cloud bursting so that applications can bring up new instances as needed in the public part of the hybrid cloud once data center capacity is reached. Application load balancing among these dynamic instances often serves as a core supporting mechanism. It can also be difficult to deterministically know where data center capacity exhausts, however, as well as how many external resources will be needed. Additionally, applications built with the capabilities to traverse public and private cloud boundaries bring about the additional challenges of ensuring that the underlying data is in the right place at the right time as well as dealing with enforcement of governance and security policies regardless of where active instances are operating.
These challenges are not insurmountable.
Capacity planning for hybrid cloud
To address data center capacity planning, load testing against a proposed infrastructure configuration, trending based on previous growth and building analytic models can help enterprises create accurate estimates of when on-premises capacity will start to struggle and require public cloud overflow processing. Of these, taking the time to throw simulated application traffic at a pre-production application environment can be one of the best ways to create awareness of where your environment is in relation to current and projected application processing needs. When properly used, this data plays a key role in helping to determine the amount of on-premises resources needed to start with and the sweet spot for the public cloud architecture. While these methods can be time-consuming and, in some cases, costly, for organizations committed to a true hybrid cloud strategy, the benefits and long-term cost savings of proactive planning as opposed to reactive re-architecture far outweigh the investment.
Hybrid cloud data location and networking
Ensuring that the correct data is in the right place at the right time can present yet another set of challenges. Enterprises may have requirements for applications to operate in various parts of their private and public cloud for resilience, scalability and flexibility. Oftentimes, applications assume close proximity to the associated data. In hybrid cloud deployments, this is not always practical, so detailed network planning to keep latency at acceptable levels between application front-ends servicing client requests and data stores hosting the underlying data must be considered.
Traditionally, this has been far easier to control with a fully in-house infrastructure, since public cloud offerings tend to provide less flexibility when it comes to networking (e.g. inability to assign more than one or a static IP address to a virtual machine instance, constraints on how virtual networks can be spanned within the infrastructure, requirements for 'internal' traffic to route 'externally' in certain scenarios, etc). Fortunately, this has been steadily improving. Pioneers such as Microsoft with their Azure offering and VMware with vCloud Hybrid Service (vCHS) are providing new dedicated, private networking solutions allowing customers to securely extend their existing data centers while keeping throughput and response levels high. These solutions can keep hybrid cloud connections from going over the public internet, resulting in higher reliability, faster speed, lower latency and improved security.
When coupled with advancements in application delivery technology that make it possible to use complex traffic steering algorithms across a fabric of private and public clouds based on business rules, organizations are now equipped with far more control and flexibility than previously possible. The key thing that organizations must keep in mind is to weigh cost versus performance and put workloads and data in the places that make the most sense for the applications, based on how and from where users typically interact with them.
Keep your hybrid cloud compliant
Compliance and privacy rules may mean keeping some data out of the public cloud, especially for organizations in verticals such as finance and medical. Just as classification, data leak protection, archiving and governance are important in on-premises data centers, they are equally if not more critical when public cloud enters the equation. To meet this challenge, cloud management platforms have come to market capable of propagating a unified set of policies across cloud borders. Vendors such as VMware, Microsoft and IBM have launched many new offerings to help companies build better private clouds with the visibility that CIOs dream about and then extend these same principles into public infrastructure. By exposing metadata from the underlying infrastructure, the applications being used and the content being accessed, and then processing these into a usable format, the challenge of compliance and governance can also be met in hybrid cloud.
What does it all mean?
These enablers all have driven the adoption of hybrid cloud strategy in the enterprise, and the outlook is positive. Modern IT has more demands than ever to provide ITaaS solutions that give enterprise lines of business increased agility, rapid provisioning and quicker time-to-market of application services. This, combined with the current gap left by an all-in public cloud model, all mean one thing: Hybrid cloud is here to stay.
Jason Dover is an expert on messaging technologies and application delivery with a background in the design and implementation of Enterprise Unified Communications and Directory solutions. Dover currently serves as part of the KEMP Technologies Product Management team. Prior to joining KEMP Technologies, Dover worked in the finance industry and provided consultative Messaging and Directory transition and migration services to NYSE Euronext and Deutsche Bank. He also served as Technical Lead for the global Directory and Messaging Operations team at AllianceBernstein.