By Michael Scott
Crypto exchange Coinbase adopts AI to avoid struggles to sustain operations during Bitcoin price crashes and surges.
In a Monday update, the US leading crypto exchange Coinbase disclosed that it is deploying an in-house-developed machine learning model to predict user traffic spikes. The model facilitates the automatic scaling of resources to avert downtime.
AI Model to Scale Resources During Traffic Spikes
Coinbase considers the AI-powered scaling of resources to increase the platform’s efficiency. Notably, the AI solution targets resolving the platform crashes witnessed during the unpredictable surge in traffic. Such plagues plague the Coinbase platform during volatile market instances.
Coinbases explains in the Monday blog that attempts to scale resources when high traffic is deemed too late to avert downtime. Instead, Brian Armstrong-led crypto exchange confirmed the development of an automatic scaling solution utilizing machine learning to predict traffic spikes.
The prior prediction will help trigger the scale-up of resources before the traffic hits the platform. Coinbase hails the model as proving its worth by averting downtime during the recent volatile spells.
Coinbase noted that as the traffic increased, its scale target doubled in a few hours to reach peak levels. The Nasdaq-listed company (COIN) added the model scaled up and down relative to the daily usage pattern. However, the process reversed when volatility declined.
The model is uniquely designed to offer a signal with an hour lead time before the traffic spike. Coinbase confirmed ditching the time-series forecasting approach relied upon to predict the traffic 60 minutes into the incident. Nonetheless, the approach proved ineffective owing to the lag time witnessed in the underlying statistics.
The Coinbase exchange platform transformed the challenge into a longer-term classification. As such, the new model utilizes external signal feeds by tracking the price fluctuations in large-cap cryptocurrencies, including Ethereum, Bitcoin, and Solana.
The model aims to indicate whether traffic is bound to surpass a certain threshold level in the subsequent hours. The approach guarantees improved accuracy.
The blog states that the critical insight in the approach is that if the model predicts crypto price volatility to increase and approach the target levels, it scales resources capable of handling the spike.
Coinbase Taps AI Algorithm to Avoid Outages
Coinbase adds that the AI model is uniquely designed to strike a balance in avoiding missed spikes and lowering false alerts.
Such is a critical step since false alerts may cause resource wastage. In contrast, excessive reduction of false alerts could lead to better preparedness. Also, inaccurate models will plunge the exchange to downtime as a direct consequence, particularly during high-traffic periods.
Coinbase has, in the past, recorded undesired experiences during market volatility. The crypto exchange has suffered several outages and technical disruptions during the crucial moments.
Recently, Coinbase suffered a significant outage on May 14, 2024. The three-hour downtime affected mobile- and desktop-based platforms. The users experienced the 503 service temporarily unavailable response on the website. Also, those utilizing the mobile application saw incorrect alerts of planned maintenance.
Coinbase had suffered a crash in March as Bitcoin strode to set its all-time high (ATH). The incident left thousands unable to profit from the gains. This awareness informs Coinbase mode to tap AI and avert crashes. Coinbase executive admitted the insufficiency in ensuring it will double capacity and resilience in its systems as users relish market excitement.
Additionally, Coinbase crashed earlier owing to the inability to handle the traffic spike during the bull run. Often, users decry the resulting financial losses due to the failure to execute trades whenever the outages occur.
Coinbase has, in the past, vowed to avert the recurrence of the issues. Its primary focus has been implementing technical upgrades blended with infrastructure enhancements.
Previously, Coinbase revealed that it was investing in scaling the server capacity. Also, it optimized the software architecture to handle the traffic volumes.
The problems behind outages emerge from traffic spikes and not steady traffic growth. As such, utilizing AI to predict unusual traffic will help automatically scale the databases to optimal levels.
Time will prove whether the AI algorithm model will help avoid outages witnessed during price swings as crypto traders flood to realize the gains in upside movement.