Why did Coinbase crash twice during Bitcoin bull runs?
Why does Coinbase crash during peak hours?
The Bitcoin community was seen rejoicing when the price of BTC reached 10,000. This was further followed by an increase, raising the price of the currency above 10K. On June 2, 2020, the main cryptocurrency registered an increase of $ 10,121. Although that price was reached after halving, it did not last long, as while Bitcoin miners have been selling much of their BTC, many Coinbase users have been seen struggling.
Cryptocurrency exchange Coinbase faced a huge reaction as it started to fall just when Bitcoin's price started to rise. It was not the first time that Coinbase has dealt with this problem. But the latest technical difficulty has left several people in the crypto community comparing this exchange to traditional banks that deteriorate during the most fragile inconvenience.
After a week of losing customers and facing repercussions from the accident, Coinbase decided to break the silence and solve the problem. The exchange revealed that it had identified the reason behind the outrage and had also fixed it. According to the blog post, when the price of BTC reached 10K on June 1st at 16:05 PDT, the exchange experienced a 5x increase in traffic in a period of just 4 minutes.
However, the company's automatic scaling failed to keep up with the significant increase in traffic.
The post also says:
“This peak in traffic affected several of our internal services, increasing latency between them. This led to the saturation of the process of the web servers responsible for our API, where the number of requests received was greater than the number of service processes, causing the requests to be queued and to time out or fail immediately.
In addition, many users were experiencing communication problems with coinbase.com, pro.coinbase.com and the exchange's mobile apps as the request error rate increased to 50%.
The integrity check also revealed that several occurrences were considered unhealthy and worsened the failure to “remove the load balancer”. In the image below, peaks are suggested to show deployments while dives are classified as unhealthy.
The company added:
“In an effort to mitigate saturation, we have reimplanted the API at 16:20 PDT to increase the machines that serve traffic. After this deployment was completed, the instances of the previous deployment were removed from rotation, resulting in another 2-minute interruption due to saturated instances and marked as unhealthy. This was handled automatically by our automatic scheduling. "
In addition, the exchange has not only claimed to have solved immediate problems, but also plans to strengthen the deployment process to ease the difficulties of automatic scaling.
