Managing Fluctuating Usage with AWS
Digicon has been working with the Baltimore City Mayor’s Office of Employment Development (MOED) on the YouthWorks system since 2011, when we developed a fully online-based solution to streamline, modernize, and generally improve its previously paper-based process. As our partnership with MOED and our work on the system has continued to evolve, we have leveraged our Amazon Web Services (AWS) Partner Network (APN) Consulting Partner status and AWS product knowledge to host all the application and reporting servers in the AWS Cloud, which provided MOED with scalability, flexibility, and reliability.
The YouthWorks system experiences a peak season annually starting with the registration period in January through April, then continuing through August for application processing and job assignments. MOED cannot afford downtime during this peak season and requires the ability to accommodate the surge of users, adjust to the scale of demand, and keep the system performing at a reasonable speed.
Digicon previously had five servers running year-round for YouthWorks, even during the off-season when only two were needed because it required costly, manual effort to remove servers after each peak season. In addition, during peak season, surges in demand required manual, time-consuming intervention in order to maintain system availability. This approach using traditional servers was an inefficient use of resources and resulted in many of the servers going unused during the off-season, or extensive downtime to address increases in demand during the peak season.
AWS’s infrastructure and services allow greater flexibility to quickly scale up and down as demand changes. Digicon uses the CloudWatch and Auto Scaling features within Elastic Compute Cloud to automatically add or remove servers as the YouthWorks system’s demand fluctuates, ensuring the user experience is consistent and stable. We also use Amazon Machine Images to standardize and streamline deployment, allowing us to spin up new instances in minutes. To ensure the uninterrupted performance and availability of the system, we use Elastic Load Balancing to distribute users’ system requests across servers, so no single server becomes a point of failure.
With the use of AWS, Digicon can now quickly scale down to one or two servers for the system in the YouthWorks off-season, then spin up more again in time to accommodate the highest demand. In a recent peak season, the system surged up to 12 servers, which was easily handled using AWS. With traditional servers, the system would have been down for a significant period of time and required procuring and configuring resources to manually expand the environment.
Results and Benefits
MOED experiences the overall benefits of the YouthWorks system being more efficient and reliable. Using Amazon’s services, Digicon has reduced downtime caused by unexpected spikes in usage or server failures, which previously required manual correction that resulted in even longer system outages. We have maintained a 99.98% uptime since hosting the system on Amazon. Using Elastic Load Balancing, we have also reduced the system’s home page loading time from 5 seconds to 0.5 seconds.
By automating the process of scaling servers with AWS, Digicon has also helped increase the YouthWorks program’s participation and success. Since we began supporting MOED to efficiently process its youth applications, the program has grown from processing a few thousand applications per season to 16,000 applications. As the program flourishes and enrollment numbers increase, the AWS system environment automatically scales up without any limitations on capacity or performance.