November 20, 2013
This article was originally posted on the SandHill blog and can be found here.
The federal government’s recent launch of Healthcare.gov has been a less-than-stellar start to say the least. It is a prime example of outsourced IT gone wrong as the multi-contractor project has been flagged with issues. It was so bad that it had to be addressed immediately by the White House, which is now taking action to fix the issues the site is currently experiencing. It is vital for enterprises to know some of the key elements in building an application/website like Healthcare.gov in order to avoid the same fate.
In order to understand the challenges the IT professionals tasked with fixing it faces, it is important to understand exactly what Healthcare.gov is. Most of the country is under the impression that Healthcare.gov is simply a website. But in reality it’s an application, accepting data, storing data, providing results based on database queries, etc. Applications need resources — like memory, CPU, bandwidth, connections to other apps, like databases, etc. That was a major issue with Healthcare.gov; the resources weren’t substantial enough to handle the load.
Whether your organization is an enterprise with a large internal project or is a service provider, here are four tips to consider when working on something similar to Healthcare.gov.
1. Provide predictive insights
In order to understand what is going on in an operating system, IT professionals must be able to provide more insight into workloads, performance and resource consumption in large scale-out environments spanning hundreds or even thousands of servers. Being able to provide a more in-depth and predictive analysis will help your organization better prepare for glitches and mishaps along the way.
2. Redefine performance
Oftentimes IT managers are focused on infrastructure utilization, but you need to analyze performance and capacity in the terms that are relevant to the organization. Use response time and latency to help your users better understand performance metrics.
3. Implement better planning
Planning and anticipating the effects of new application rollouts is important. It might not be practical to test with production-level systems and loads, in which case it’s crucial that the planned architecture is analyzed using methods that can predict how it will perform. That way you aren’t leaving the success of your rollout to chance.
4. Make time for testing
Even if you’ve used predictive analytics to plan an architecture that will support expected workloads, you still need to test. If requirements change during the test phase, it’s not a good idea to skip testing in order to make your target launch date. Proper testing can mean the difference between knowing your real capacity requirements before production and spending more resources on server capacity after production. Better to reconfirm and succeed than launch on time and fail miserably.
The disastrous failure of Healthcare.gov does not have to be the same fate for your company. Taking the time to properly prepare is key; and if you start by taking these tips, you will be on your way to a much more successful launch.