There are many changes underway in networking architectures, driven by increasing cloud adoption, the trend towards microservices based applications, and software eating the world. In this context, many people are uncertain about future of Application Delivery Controllers. Here is my view.
The Load Balancing Era
The internet grew at a dramatic pace by adopting simple, but scalable techniques. Web scale architectures delivered tremendous availability and scalability by embracing an approach of horizontal scale out. In the mid-1990’s, Load Balancers became an integral part of web scale architectures, providing server monitoring and load balancing.
The humble Load Balancer appliance came to occupy a strategic piece of real estate in Data Center architectures. As HTTPS came into use, at least for important/transactional traffic, SSL offload became an important addition. All traffic from the internet flowed through load balancing appliances, so it became an insertion point for many functions of value to the web site: HTTP/TCP optimization, in-memory caching, sslvpn, web application firewall, and much more. This consolidated approach eliminated appliance sprawl, and provided consistent management of networking functions. And the vehicle for such “aggregation” was given the name of Application Delivery Controller.
Highlighting this transition, in 2009 Industry Analyst Firm, Gartner, published “Load Balancers Are Dead: Time to Focus on Application Delivery”. This did not mean that load balancing functionality was dead; rather, this functionality became strategically important as the anchor tenant in the more valuable ADC.
The Application Delivery Controller Era
There was a decade of growth of the ADC market, along with consolidation towards the stronger vendors. During this period, there started to be many other important changes in the networking landscape. Among these were:
- Networking packaging expanded beyond appliances to virtual appliances and software
- On-premises to cloud and hybrid deployments. In cloud deployments, it less desirable (and possibly undesirable) to consolidate functions.
- The growth of microservices based application architectures
So, what’s next?
My view. We will see a period of “disaggregation”. The functions, including load balancing, which were aggregated into the ADC will take on life as distinct functions again. However, this will not be a simple reversal of the earlier process of “aggregation”. Rather, it will be important to the have “composition” of these disaggregated functions. Important aspects of composition will be coherent management/control of disparate functions or services, and the ability to chain functions/services together under the aegis of a common policy framework.
These network functions will be composed for different situations. For example, there will be cloud services which provide application access with security. The “Service Mesh” will take these functions deep into the fabric of microservices bases applications.
The term “ADC “might wither, but do not be fooled. This does not mean that the functionality of the ADC is less valuable; rather, these functions will be become the core of new services architectures.