IT infrastructures are evolving to a virtualized model through cloud, hybrid cloud and software-defined data centers, a shift designed to create a more dynamic and flexible application delivery architecture. But agile networks that can provide on-demand availability seem to be contrary to traditional, fixed networks, which provide a more stable framework for IT functions. Is it possible to have the best of both worlds?

Hardware or software?

Traditional networks are built using highly available technologies and designs delivering a fault-tolerant and resilient infrastructure. Routing protocols along with active/standby and active/active designs create an environment that ensures application availability under consistent demand conditions. The components are typically based on proprietary, vendor-designed hardware, which makes it difficult to make significant and ad hoc changes.

As the cloud emerged, businesses started leveraging the ability to move their applications quickly and easily. This continued to evolve as virtualization technologies abstracted the proprietary hardware platforms by using software-based functions on common-off-the-shelf hardware instead of the proprietary hardware. The ability to provide any IT function on a shared hardware framework enables agility, which is necessary to staying relevant and up to speed in this digital age.

The introduction of orchestration and automation systems for virtualized architectures can quickly and dynamically alter the infrastructure’s application delivery characteristics.  Automated agility leads to elasticity, which is the on-demand resourcing we desire.

But at the same time, a dynamic agile and elastic architecture inherently provides more opportunities to misconfigure or break the delivery of the application. Each time the environment is adjusted is an opportunity for a change to introduce a problem.

The evolution of IT architectures to these virtualized designs cannot succeed if we replace stability with agility, since the flexibility of the application delivery infrastructure doesn’t matter without stability.  Having the fastest car is irrelevant if it only runs 10 percent – or even 90 percent – of the time. We need technologies and functions that can ensure the stability of these dynamic software-based architectures.

Mature technologies in new roles

Thankfully, those technologies already exist. Before virtualization, we developed technologies to abstract the application servers and application delivery infrastructure. Server load balancing and global server load balancing technologies through application delivery controllers provide scalability and reliability by clustering similar functions and application monitoring. 

Application delivery technologies are critical to enabling the flexibility of the virtualized networks because they enable agility and elasticity through the abstraction of the application servers and physical location of the services.

These ADC (News - Alert) technologies also provide application monitoring, which means they can ensure the stability of the end-to-end application delivery from client to server. When an individual application server fails, the server load balancing function can automatically detect and mitigate the failure. If the problem is location- or network-centric, global server load balancing technologies can redirect the application traffic to available resources.

For the ADCs to deliver maximum value to the virtualized architectures, they need to monitor and understand the metrics of all of the components in the application delivery path. This includes the servers, data center environment, LAN, WAN and even the client performance. Application performance monitoring tools need to integrate the ADC metrics as a key component in their analytical process.

The last piece of the solution is to integrate orchestration and automation technologies to automate the dynamic processes. Orchestration and automation enable near real-time elasticity and fault mitigation. Manual processes and operator intervention aren’t realistic – no one can make changes to the dynamic virtualized network at the frequency required.

Dynamism changes and fixes application delivery

Today’s IT organizations are focusing on applications and application delivery, but the virtualized network is becoming an abstracted infrastructure that makes up a framework for application delivery. IT operations are increasingly less network driven and must take into account application functionality and SLAs, as well as have the ability to monitor and manage the IT architecture focusing on the assurance of application access.

The adaptation of existing mature reliability technologies for the virtualized cloud, SDDC, and software-defined architectures enables businesses to get the benefits of agility and stability through the same workflow processes.

Frank Yue (News - Alert) is director of application delivery solutions at Radware (www.radware.com).