If You’re on the Highway to the Cloud, Make Sure You Pack the Right ADC to Take You There


After recently reading about VMware’s view on the evolution of IT towards cloud services, or as they call it the “journey to the cloud,” I would like to share a few insights about how application delivery controllers (ADC) should fit in this journey.

The whole idea in transforming traditional IT services into a self provisioning cloud is built on two foundations:

  1. Virtualizing the infrastructure that provides IT applications and services
  2. Automating the processes that take place every time a new application is rolled out or an existing application is updated (e.g. moved to a new server, needs to provide more capacity, etc.)

It’s impractical to provide process automation without first virtualizing the IT infrastructure. As most applications are serving business critical processes, an ADC service often goes hand in hand with the deployment of the application. So, the virtualization evolution should also cover the ADC part of the infrastructure, to enable an end-to-end cloud solution with fully automated services.

But how can you virtualize the ADC layer in the network? It is often built on a dedicated appliance, just like physical servers were, before VMware provided a solution to transform them to virtual machines. Using soft ADCs running as virtual machines on top of VMware’s virtual infrastructure (as an example) is indeed one possible solution to this.

The limitation of this solution is performance and its predictability: It’s no coincidence that the majority of ADCs today, are running on dedicated appliances – one of the main reasons for using an ADC is to improve application performance. To achieve that, ADCs are running on hardware optimized machines, with hardware accelerators for ADC services such as caching, compression, SSL offloading etc.

Another possible solution we’ve started seeing in the market, is a new type of ADC appliances with a hypervisor layer on top of it, which can run several virtual ADC instances (vADC) on a single device. Such a solution can bring similar benefits to those of server virtualization, like consolidation of multiple physical ADCs into one device, a simplified infrastructure that directly drives a solution’s cost down.

In reality, we see that the solution is the combination of both virtual ADC form factors – soft ADCs (often used for application development) and vADCs running on hardware optimized appliance (for production environment).

It is only by using a virtualized ADC infrastructure with process automation that enables self service IT tools, that this can be achieved. Take the VMware model for the cloud as an example. VMware adds to its server virtualization layer, new provisioning, management and process automation tools, and through them forms packages of virtual applications, which can be provisioned and managed in a click of a button.

Adding to those application packages with ADC services is only possible if the ADC computing resources are also virtualized, and offer similar services to those offered by the VMware solution, while integrating with their process automation tools interfaces (e.g. using dedicated plug-ins).

So what is my insight here? If you are planning to shift your traditional IT solution model to a cloud type of service (tomorrow or somewhere in the next couple of years), you should start thinking about your ADC infrastructure the same way you think about your servers: They both need to be virtualized, they both need to provide process automation capabilities and it would be best if those automation processes could interoperate, to enable real transformation of your IT infrastructure into a self service cloud.


Please enter your comment!
Please enter your name here