Optimizing your app server performance really matters

That's why a intuitively load balanced and scalable server can mean sink or float for many business on their rise. Because of this, we decided to sit down and have a chat with Ranga Rajagopalan, the CTO of Avi Networks.

ADM: Tell us a little bit about Avi Networks?

Rajagopalan: Thank you for this opportunity to discuss Avi Networks and our exciting announcement. Avi Networks delivers agile application services beyond load balancing including deep application analytics, predictive autoscaling, and security in the data center or public cloud. We do this with a load balancing platform built on software-defined principles that can run on standard x86 servers, VMs, or containers in a cloud-independent framework.

Enterprises want public-cloud-like agility but traditional appliance-based load balancers can’t solve the challenges of cloud-native applications in modern data centers and can’t elastically scale to meet specific traffic loads. We built Avi Networks to match the application service needs of these enterprises and their dynamic environments.

ADM: You just announced breakthrough scalability using Avi Networks application delivery controllers, can you tell me a little bit more about that?

Rajagopalan: We demonstrated the scalability of Avi’s Elastic Service Fabric by scaling up and scaling down from zero to one million SSL transactions per second with no impact to performance. This represents a new industry milestone for automated, elastic load balancing and application delivery. It addresses the needs of large retailers, financial services, technology, and service provider enterprises struggling to scale up or down in response to seasonal traffic changes, fend off DDoS attacks, and deploy hybrid cloud applications in a cost-effective manner.

ADM: How is this any different from what load balancer vendors currently offer?

Rajagopalan: With Avi Networks’ software-defined load balancers, IT departments can pay for maximum load balancing capacity on an as-needed basis, and scale it down as traffic returns to normal levels. Legacy architectures and hardware-centric load balancers are proving to be inflexible when it comes to cost-effective scaling and handling modern application architectures driven by automation, hybrid-cloud requirements, and programmable self-service. They require huge investments in capital equipment that goes unused during normal periods of traffic. They have one way scalability – they can only scale up whereas Avi Networks can elastically scale software-defined load balancers on x86 commodity hardware up and down, and in any cloud to meet specific traffic loads. Another important capability that is built-in with Avi Networks is the deep application performance insights that are delivered by the platform which reduce network troubleshooting times to seconds.

ADM: Does this have an impact on the price point for ADCs?

Rajagopalan: This demonstration proves that Avi’s load balancers, built using a software-defined architecture, are able to deliver elastic application services and can save costs for enterprises. These enterprises find that they spend significant amounts on load balancers built with proprietary hardware which are overprovisioned and work only in their data center. They often have to find a different solution with a different architecture for load balancing services in the cloud.

The Avi Networks demo was performed on the Google Cloud Platform but can easily be replicated on-premises or in any other cloud. It ran on 40 servers at an infrastructure cost of just $1.20 per server per hour, scaling out to handle one million SSL transactions per second and scaling back just as easily and efficiently. Handling similar transactional capacity with traditional hardware load balancers such as F5 Networks or Citrix NetScaler would cost millions of dollars with no way to scale back down when traffic returns to normal.

ADM: What is the core enabling technology behind software-defined ADCs that makes it more effective than hardware appliances?

Rajagopalan: Avi Networks uses software-defined principles to bring public cloud-like simplicity and flexibility to application services including load balancing, autoscaling, application visibility and analytics with complete REST API-driven automation. We have created an elastic service fabric consisting of distributed load balancers that can be managed as one. The central control plane in Avi Networks is a single point of management, control, and orchestration for the distributed data plane consisting of software load balancers. These distributed load balancers deliver real time telemetry about applications, security, and end users to the controller which analyzes the data and displays rich insights to busy administrators, significantly simplifying their troubleshooting efforts. Avi is the first solution that uses analytics to drive intelligent decisions on traffic management and elasticity based on traffic or throughput thresholds.

ADM: Do you see a customer need for this kind of scalability?

Rajagopalan: Avi demonstrated its scalability in response to a real customer need. The customer is one of the world’s largest retail services and processing company that requires massive elasticity and automation. The significant rise in the amount of TLS/SSL traffic has placed an increasing burden on load balancers. Customers like this one find that with Avi they can:

- Eliminate the over-provisioning that is typical of expensive hardware load balancers (can only scale up and then they are stuck with that capacity). In many cases, achieving this scalability is not even possible with legacy load balancers.

- Grow and shrink their application services in response to their real business needs.

- Address hybrid cloud use cases, seamlessly expanding capacity from their data center to the public cloud due to Avi’s intelligent traffic management across clouds.

- Handle DDoS attacks by simply scaling their application services, ensuring good performance for legitimate users while investigating and blocking bad actors/domains.

ADM: You mentioned SSL traffic, how do you see that affecting application performance?

Rajagopalan: The Internet is seeing ever increasing volumes of encrypted traffic in response to the higher incidence of cyber-attacks - over 50 percent of all Internet traffic, including Netflix movies, is now encrypted with SSL/TLS. This presents new challenges for network managers and application architects who must ensure that their applications are secure and responsive for end users. The growing amount of encrypted traffic puts enormous strain on load balancers that must offload or decrypt these SSL-encrypted packets before they reach the application. The purpose of our demonstration was to prove how a software-defined architecture for load balancing delivers the most elastic and cost-effective approach for the growing volume of encrypted SSL traffic.

ADM: Other than scalability, any other trends you see in the market?

Rajagopalan: With companies rapidly managing their business and IT infrastructure with software-driven solutions and processes, application owners will have a greater say in IT purchasing and resource allocation to deal with issues including APIs, programmability and shadow IT. Similarly, infrastructure and operations teams will require greater visibility and troubleshooting tools to accelerate delivering apps from hybrid clouds and to optimize app performance.