Inspired by mathematical finance, an applied mathematics field that analyzes the stock market by Microsoft and MIT, they have created a mathematical model that improves cloud computing utilization while maintaining availability by improving cloud service traffic efficiency.
Cloud service is a network business that provides various computing services to customers by connecting networks with optical fiber lines between data centers set in various regional cities. One of the key points in cloud services is availability, which is the percentage of time the system can use.
When trying to use a service with low availability, it may not be available. However, in large-scale computing systems, the system itself may be down due to unexpected accidents such as power outages or line cuts, so network and data center self-backups are required to increase availability. But backup networks and data centers are usually useless without problems. Increasing the backup will lower the overall utilization rate. Of course, the decline in the utilization rate comes back to cost.
Using network data from Microsoft, Google, IBM, and AT&T, Microsoft and the MIT research team mapped the probability distribution of traffic outgoing from the data center, calculated the appropriate bandwidth for each link, and developed a mathematical model to predict the occurrence of failure. This mathematical model applies the existing mathematical model studied in mathematical finance to network traffic by comparing traffic usage to stock market investment and network failure to loss.
As a result of experimenting with this theory in an environment simulator close to reality, it is said that it has been confirmed that traffic disturbances are prevented while maximizing the amount of transmission. According to the research team, the published model can increase the amount of data transfer by three times that of conventional network designs without incurring data loss.
According to the published mathematical model, the research team explained that if the network utilization is increased, the cost of using the cloud service is reduced by reducing the burden of the provider, and it is possible to reduce enormous energy consumption by the cloud infrastructure, which can have a good impact on the environment. Related information can be found here .