A few days ago, I launched the “Load Balancing” Node: https://steem.justyy.workers.dev

It is a Javascript code that runs on the CloudFlare edges (more than 200 network centers). The network is the computing. When requests are received, the Node will send a ‘ping’ request to a few nodes, and whoever returns first win the request.

See this on the CloudFlare worker debugging console – the remote virtual debugger.

But all those work are behind the scenes. The user will not see these details, and the Load Balancer is just working fine – and it indicates the actual Node that served the requests via the Response’ custom header `origin`

It is a perfect use case to let the CloudFlare workers handle the Load Balancing and route the traffic to the fastest server. The serverless technology is cost-effective (no need to pay for the servers), secure (patches are applied/installed automatically), hassle-free (you don’t manage the servers) and fast i.e. the Workers run on the Cloudflare Networks (more than 100 countries and 200 network centers – usually mean shorter pings).

Free plans are limited to 10ms CPU time per worker script, 30 workers, and 100K requests per day. Worker unlimited are limited to 50ms, and 10 Million requests per month (always shortest latency to worker nodes)

–EOF (The Ultimate Computing & Technology Blog) —

推荐阅读：
没有捷径可走　 计算机向π值的挑战　 爱因斯坦与数学速算　 悖论破解法　 啊哈反应　 圆形相遇和追及问题　 抽屉原则问题　 利用抽屉原则解题——制造抽屉　 小学画展题　 乘积最大问题