Application Services Must Follow Apps as They Follow Data

F5 Miniatura
Published September 25, 2018


The adoption of cloud and now containers has disrupted the traditional stovepipe network architecture that has dominated data center design for decades. Control over traffic is no longer confined to well-defined locations in a single, network architecture. It is now distributed across the Internet and throughout the data center as containers abrogate increasing capabilities from application delivery.

That distribution of applications is wreaking havoc with operations as visibility, security, and performance of applications and supporting infrastructure can no longer be assured through control of the data path.

It is clear that the traditional strategic control point of the network is being disrupted, and with it the application services that have long assumed its existence.

For years, these application services have been delivered on platforms known as Application Delivery Controllers. The ADC, in turn, has relied upon hardware for the scale and speed needed to match the performance and availability requirements of the business. The ADC as a hardware solution has been reified by the requirement to handle the load of hundreds and thousands of applications in a single device. This choice has been dictated by applications, which are tethered by the data for which they are the primary interface.

But application services are not bound to that hardware, because they are and always have been software. The benefit of purpose-built hardware has always been its ability to improve performance and scale, providing orders of magnitude increases over commodity hardware.

F5 Major Milestones

Just as application services are not bound to hardware, neither are applications bound to the data center. Applications are, in fact, more tightly bound to their data. When that data moves, so do the applications. This cause and effect relationship is critical to the future of application delivery because we are on the verge of generating more data than we can move. According to Cisco's "The Zettabyte Era: Trends and Analysis", by 2021 we will generate 60 zettabytes (ZB) of data. Our capacity to move that data, however, remains much lower at 3.3 ZB. The gravity of data will necessarily tether applications to the environment in which that data resides. That environment may be the data center, but it may also be in the cloud.

This coupling of applications to their data is also seen in a similar coupling of application services to applications. As applications move, so must their application services. This has an impact on the data path.

Architecture and operating models have evolved to accommodate a predictable data path, but that predictability is now gone. The services that make up cloud-based applications follow no prescribed interconnectivity.

Networked services must transform.

Gone is the single, well-defined data path and with it the strategic points of control at which application services have traditionally been deployed. The data path is now variable, dynamic, and distributed. There are more of them, as cloud and business adopt per-application approaches to development, deployment, and delivery. Application services can now be inserted and executed anywhere along that path.


Service Model Comparison

To follow the applications, application delivery needs to adopt a model for application services that is not confined to the data center network. Instead, it needs to embrace alternative deployment locations and operating models that include cloud-native and service-based offerings. 

This approach introduces new options for the insertion and delivery of application services. Application services might be inserted into the application or client but execute within the confines of a cloud or as a service. They might be built into the application as binary components or included with meta-data such as tags. The location becomes less important than the ability to follow the applications as they follow their data.

Options are Opportunities

It may be tempting to view these options as a threat, as they are not conducive to deployment of a traditional ADC.  They are inherently disruptive to traditional delivery architectures and force both vendors and application delivery professionals to expand their domains beyond the comfortable confines of the network.

But it is better to see these options as the opportunities they are. These new options are opportunities to embrace new application services and operating models as a way to continue to deliver application services to make sure apps are secure, fast, and available.