Today’s guest post comes from Ephraim Baron, an IT operations expert at Equinix, a global platform of high-performance data centers housing many of the world’s largest public clouds and a RightScale partner.
Application performance and end-user quality of experience (QoE) are critical to IT success. The primary goal of information technology is to make users – and, by extension, enterprises – more productive. This fact makes end users the ultimate arbiters of IT value and why it’s so important for IT to focus on user experience. End users care about utility and ease of use. If an application or service doesn’t add value, users won’t embrace it. Similarly, users will let it be known if an application is not easy to use.
There are two main ways in which an application can be difficult to use. First, it may have a poorly designed or non-intuitive user interface. That’s an issue for developers and is outside of the scope of this article. The other is poor performance, and application performance is IT’s responsibility. So what can IT do to ensure good application performance and, by extension, happy users? I’ll explore the concept of service radius and offer ideas for designing, deploying, and managing a high-performance application delivery platform.
The Evolution of Application Delivery
Before I dive in, though, let’s briefly look at application architecture. Almost all applications have at least two parts: the user front-end and the data back-end. For single-user applications such as word processing, spreadsheets, media viewers, and the like, both front and back ends may run on the same device. For many others – particularly enterprise applications shared by many users – front and back-end systems are separate. Users connect to these applications over a network using a client interface. The network exists to serve the application and ultimately the end user. If the network isn’t matched to the application requirements, performance suffers. This makes network bandwidth and latency important determinants of user experience.
In the traditional application delivery model, all back-end systems are hosted from a single, central location – perhaps a headquarters data center. For people working at headquarters, the experience is good. But the farther away users are, the worse their quality of experience gets. The trend toward any time, anywhere, on any device (ATAWAD) computing increases this effect.
As enterprises become more global, mobile, and virtualized, the central model for application delivery is becoming less and less viable. End-user quality of experience is tied closely to response time, which in turn depends on connectivity and proximity. The combination of application requirements and distance is a major determinant of end-user satisfaction.
This brings us to the concept of service radius. It defines how far away users can be from their applications and still get good performance, and it’s typically expressed in terms of network latency. Some particularly sensitive applications – streaming video, for instance – might have a latency target of less than 10 milliseconds. More tolerant apps may continue to perform well up to 100ms. The key is to understand application characteristics so you can define a latency threshold. This information is then used to design a distributed application delivery platform that locates application resources at appropriate proximity to end users.
Re-Architecting for Performance
The design of a performant application delivery platform involves five basic steps:
- Identify and Locate Your Users: You need to know where users are located before you can decide how best to serve them. User populations may fall into a number of classes, such as major sites, satellite offices, and field-based workers as well as customers and partners.
- Establish Application Requirements: Next, you should determine the performance characteristics of the applications your users depend on. By understanding how your applications perform, you can establish where your systems must be to satisfy user needs.
- Optimize Your Network: The next step is to map out network performance hubs (NPHs). These are regionally distributed installations of equipment delivering data, voice, video, and related network services to enterprise employees, customers, and supply chain partners. The location of NPHs defines the service radius at which applications can be delivered within performance requirements defined in step 2. This is where a global, carrier-neutral hosting provider like Equinix can provide a lot of value.
- Distribute Your Applications: This is the most challenging step and will likely require a collaborative effort between enterprise architects and the network team. Together, they will map data flows and determine how best to divide and/or replicate data to meet user needs.
- Optimize Service Consumption: As you distribute your application architecture, you’ll likely want to consider a variety of IT delivery methods – including public and private clouds, managed service providers, and SaaS-based vendors. RightScale Cloud Management, with its integrated service platform, can greatly assist with the orchestration and ongoing management of these resources.
The Evolution of IT
The days of IT’s absolute authority over users are drawing to a close. IT is increasingly being treated as a business, with customer satisfaction as a primary goal. Customers like developers and application owners are finding it increasingly easy to bypass traditional IT when their needs are not being met. The good news is IT can remain relevant by focusing on user QoE. In addition, as IT operations transforms from a systems-based to a services-based focus, they move further up the value stack. Rather than being viewed simply as overhead – a cost of doing business – IT can establish itself as a strategic enabler and valued business partner.
It truly pays to keep your users happy.
To learn more about the joint RightScale and Equinix solution, see how we can help you build the hybrid cloud environment you need.