Everywhere you turn it seems someone is talking about the “cloud.” It’s much like the e-business craze of the late 1990s/early 2000s. Everyone had an e-business solution, or at least said so even if it was just for marketing purposes. Now, everyone is building cloud solutions.
Although the term is certainly overused, the value and potential that this utility computing model offers is hard to overlook. When you think about what customers—whether internal IT shops or cloud service providers (CSPs)—are really looking for, it boils down to two main things: increased agility and cost savings. Agility comes in the form of rapid procurement and provisioning as well as the ability to elastically grow and shrink services on demand. On the cost side of the equation, it is the ability to leverage commodity hardware, avoiding the expense of costly hardware appliances, and the ability for customers to pay for what they use rather than buying products that meet the high end of their capacity planning estimates, which typically factor in their just-in-case scenarios.
For CSPs, who are much further along in their cloud strategy than most enterprises, these elements are critical to their business models. In fact, if you gave them a dollar to invest in either of these cloud features, or standard service features, they would likely put 70 cents toward the ability to turn up a service on demand, grow and shrink it as needed and pay for what they use. The remaining 30 cents would go toward service features and would be tied to what they actually would be using, not a laundry list of features that look good on a data sheet but aren’t relevant to day-to-day activity.
Server virtualization has demonstrated that it is possible to achieve new levels of agility, simplicity and efficiency by decoupling application software from the compute hardware on which it operates. Storage architectures have been evolving to be similarly decoupled from the application and the hardware. But is it possible to do the same to the network?
The answer is yes, and progress is being made on this front, but it is still lagging behind because it represents one of the biggest challenges in building elastic, enterprise-class cloud computing offerings. Unfortunately, because the network isn’t agile, it is often the bottleneck for cloud adoption. It is not easy to make Layer 4/7 network services such as server load balancing, VPNs/firewalls, IDS/IPS and so on less hardware-dependent while maintaining the ability to scale out.
As a result of limitations on the networking side of the infrastructure, deployment models that CSPs have at their disposal for offering virtual network services pose a real challenge. You see, CSPs essentially have two options: an over-the-cloud approach and an under-the-cloud model.
The more you talk with those building infrastructure-as-a-service (IaaS) offerings, the more it’s becoming clear which approach they should take: under the cloud, not over it. What does this mean? Let’s look at each one.
Over the Cloud
From a networking perspective, this deployment option has evolved as a result of limitations in current hardware and virtual appliance products on the market.
Why call it over the cloud? Because this model isn’t purpose-built for the cloud. You essentially take some functions and bolt them on over the cloud infrastructure you can offer (compute and storage).
In an over-the-cloud scenario, the only infrastructure that gets truly offered as a service to end users is a pool of compute capacity, generally implemented as a set of orchestrated virtual machines (VMs), that you then let the cloud user figure out how to use. Basically, users can load some application workloads on some of these VMs, or they could potentially use some of the VMs for network services, typically in the form of a virtual appliance. Some cloud providers have tried to work around the limitations of the over-the-cloud model by selling pre-provisioned virtual appliances—already pre-loaded—but there is still complexity, as the end users must do the configuration. Despite this work-around, the reality is the only service (i.e., value) that the CSP is adding is the initial installation of the software on the VM.
If that’s not enough, with an over-the-cloud model, end users also must deal with independent, individual virtual appliances for each service (a load balancer from vendor X, a firewall from vendor Y and so on) AND learn how to configure and use each one of them.
A CSP offering over-the-cloud services ends up putting the complexity square in the lap of the customer. The customer is responsible for choosing, installing, configuring, operating, managing and maintaining these network services. This creates a situation where both the customer and the CSP lose, as there is little value creation and thus no business opportunity.
Under the Cloud
This should be the preferred approach for CSPs, as it is much more optimized for the cloud and IaaS. It is where the delivery of network services has been redesigned to clearly provide the cloud function with all of the agility, dynamic provisioning, elasticity, scale-out performance, multi-tenancy and cost savings the cloud promises.
The CSP takes ownership of the complexity that is handled by the customer in the over-the-cloud model. It delivers the network service inside their infrastructure, and what it exposes to the customer is not the product, but rather the function, which it then delivers as a service to a customer. All of the sudden, the customer can order a service, such as server load balancing, and have it procured and provisioned in minutes (essentially on demand), without having to worry about the complexity of implementing the service itself. It is all under one roof, or the cloud.
Using the under-the-cloud approach, CSPs also return to their value proposition roots where they deliver a service so their customers don’t have to worry about it. It lets cloud customers focus on providing value to their respective business and not dealing with the headaches of managing the infrastructure. With an under-the-cloud service offering, CSPs have laid the foundation for true network agility and cost savings that they can pass on to customers. They can now offer more true cloud services, they differentiate their businesses, and they can reduce customer churn.
The bottom line is this: The cloud changes everything. It requires new ways to deliver services and a new mindset for providers of services and users of them. In this new world it is better to go under than over.
About the Author
John Vincenzo is the vice president of marketing at Embrane, where he is responsible for the global promotion, positioning and overall marketing of the company’s virtualized network services solutions. John has nearly 20 years of strategic high-tech marketing experience, including in the networking and telecom industries.