Google Shows How To Scale Apps From Zero To A Million Requests Per Second, For $10

In the world of web apps, how fast your app can scale can mean all the difference. In what appears to be the latest barrage in the Google versus Amazon battle for cloud supremacy, the company demonstrated an interesting method to scale to 1 million requests a second for the astonishing low price of just $10.

Using its publicly available Google Cloud Platform services,Anthony F. Voellm, a Google Performance Engineer,was able to demonstrate how to scale a cloud application deployment within 5 seconds after the setup, without pre-warming, enabling Google Cloud customers to serve 1 million requests per second and sustain that level for an indefinite period of time.

To put this achievement into context, 1 million requests per second is 20 times greater than the throughput in last year’s Euro vision Song Contest, which served 125 million users in Europe and was also hosted on the Google Compute Engine. According to the post by Voellm, “Starting with an empty Compute Engine project and ending with 456 cores provisioned and one load-balanced IP address actively processing 1.016M Requests Per Second (+-0.007M) took a total of 7 minutes 30 seconds.

”“This setup demonstrated a couple of features, including scaling of the Compute Engine Load Balancing, use of different machine types and rapid provisioning. The load was directed at a single IP address, which then fanned out to the web servers.“ Said Voellm.

In a tweet,Lydia Leong, Research VP at Gartner, who covers cloud computing, data centers, and Internet infrastructure explains the importance of the milestone, saying,  “Relevance of GCE LB load test: With AWS ELB, if you expect a big spike load, must contact AWS support to have ELB pre-warmed to handle load.” Shegoes on to explain”It’s commonly thought that what AWS is doing on the back-end with pre-warming is scaling out HA proxy instances for your ELB, or the like.

”Pre-warming is the concept of turning on web servers before the capacity is required, rather then spinning up the servers as demand requires. HA Proxy is a free, very fast and reliable solution offering high availability, load balancing, and proxying for TCP and HTTP-based applications.Amazon Web Service also offers a similarElastic Load Balancingservice but unlike Google, requires a pre-defined starting point for its initial capacity, scaling up or down based on traffic.

According to the Amazon Web Services site,upon request, the Amazon elastic load balancer can be configured to have a minimum scaling based on the traffic that is expected. This is needed when flash traffic is planned, when a DDOS attack is initiated against the load balancer (or the customer’s instances), or when the load test cannot be configured to gradually increase traffic.The key difference is that Amazon requires a premium support agreement and users make a request through its support team to pre-warm the load balancer. Google is saying that anyone can instantly scale, with no expensive support contracts required.