A10's Harmony brings app delivery services to a multicloud deployment
As use of cloud services explodes, A10 Networks is working to add application delivery services to the multicloud deployment process.
The use of cloud services has exploded over the past five years. ZK Research forecasts cloud services will grow at an 18% compounded annual growth rate, or CAGR, which is six times greater than rest of IT. While the cloud is referred to as a singular entity, the reality is most organizations will operate hybrid cloud environments, which will likely include multiple public cloud providers. Another data point from ZK Research supporting this statement is 82% of organizations have plans to implement hybrid clouds, so this multicloud deployment model is rapidly becoming the norm.
The shift to hybrid cloud raises an interesting question: How do IT operations teams deploy applications services to an environment increasingly dynamic and distributed? The beauty of the cloud is workloads can be spun up and down on demand. Also, the decoupling of the application from the underlying infrastructure enables software to be migrated from the private cloud to the public cloud, or between cloud providers. Alternatively, microservices enable parts of applications to run in different clouds for regional or cost optimization.
This necessitates the need to be able to deploy application services, such as load balancing, Secure Sockets Layer inspection and network address translation, in an equally agile way. The problem is application delivery controllers (ADCs) enable the services to be deployed on a per-location basis, but the cloud requires them to be delivered dynamically on a per-application basis. This means if an application is spun up in a container, the ADC service needs to be spun up equally fast.
Challenge for IT operations teams
The problem for IT operations teams tasked with multicloud deployment is traditional ADCs don't have the necessary agility to make this happen. In fact, most data center infrastructure is designed for per-location deployment. What's needed now is infrastructure with the same level of dynamism and agility as the cloud.
A good example of this is A10 Networks' recently announced Harmony Controller. Harmony is actually much more than an ADC; it is a cloud management, orchestration and analytics engine that is an extension of the Lightning Application Delivery Service for cloud applications A10 launched last year following the company's 2016 acquisition of Appcito.
Harmony Controller was built cloud-natively to be microservice- and container-based. This is significantly different than the approach Citrix took with its CPX product, which was to take its virtual NetScaler ADC product and run it in a container.
This is a critical point to understand. A cloud-optimized ADC enables specific services to be run in containers and be spun up anywhere in the hybrid cloud environment. This is markedly different than running the entire ADC stack in a container. In addition to providing a multi-tenant ADC service, A10's Harmony Controller also incorporates real-time analytics on a per-application level and centrally manages the orchestration of app services in hybrid clouds.
Approach similar to SDN controller
The easiest way to think about Harmony is it is to application services what the software-defined networking controller is to network services. The per-app analytics give operations teams a single point of visibility and intelligence to invoke application services, capacity plan and optimize the infrastructure that could be located in the company's private cloud or any number of public cloud services.
At launch, A10 announced support for VMware-powered private clouds, Google Cloud Platform, Amazon Web Services and Microsoft Azure.
The cloud is bringing an unprecedented level of agility and elasticity to applications, and ADCs need to evolve to enable businesses to deploy application services at cloud speed. A10's Harmony is an excellent example of how infrastructure needs to evolve for multicloud deployment as the cloud becomes the dominant compute model.