Scalability is the ability of a web application to keep operating as traffic and workload grow. For an AI application, plan beyond the app server: infrastructure capacity, resource allocation, load balancing, databases, and external services all affect whether a traffic spike becomes downtime. section 1.2.1 section 9.6.1
Choose production measures according to expected load:
- Below 10,000 requests per day, establish uptime and response-time monitoring, error tracking, essential security, daily backups, and a way to add application instances. section 9.6.2
- At 10,000–100,000 requests per day, strengthen the infrastructure; a CDN can distribute static assets and reduce geographic latency. Check whether the hosting provider already supplies one. section 9.6.2
- Configure production databases and services such as authentication and rate limiting alongside the application. Vercel offers a streamlined Next.js deployment path; growing demands may eventually require moving some components elsewhere. section 9.6.1
Document-processing workload
In a RAG application that accepts uploaded PDFs and DOCX files, parsing documents and creating vector embeddings can be heavy work. Moving those tasks to background workers or serverless functions can improve responsiveness, scalability, and fault tolerance. Rate limits and upload quotas also help restrict excessive upload attempts. section 11.2.2 section 11.4
