3.3.4. Kubernetes
Kubernets is an open source-based management system that provides
automatic distribution and scaling of containerized applications. It was
designed by Google in 2014 and is now managed by the Linux Foundation. The
purpose is to provide a platform for automating the deployment, scaling, and
operation of application containers between hosts in
- 13 -
multiple clusters. Kubernets manages the belly's resources in a
form called an 'object'. Objects can serve a variety of roles by grouping a
kind of container, a set of basic containers called Pods, and even the
controller Replica Set, which manages the set of containers, and even the
Service Account and Node, can be used as one object. The role of the Kubernets
Node is largely divided into master and worker. Master is responsible for
managing the cluster so that Kubernets can function properly, and Worker is
created an application container. Master Node is responsible for all services
for deploying applications by running API Server, Controller Manager,
Scheduler, DNS Server, etc.
4. CONCLUSION
Research so far shows that the new coronavirus is affected by
temperature and humidity factors. Also, in common sense, cities with similar
latitudes have similar climates. In 'Temperature and Latitude Analysis to
Predict Potential Spread and Seasonality for COVID19', the COVID-19 virus
assumed that the transmission of the virus would be active in large cities with
an average temperature of 5 to 11 degrees per day and humidity of 47 to 79
percent [2]. Therefore, the hypothesis will be verified using actual climate
data. We will then implement a model that predicts the date on which COVID-19's
spread rate decreases. In order to verify these tests, we will make a deep
learning model that uses historical weather data to predict this year's
weather. For weather forecasting models using deep learning through `Weather
forecasting model using artificial neural network' [13], the deep learning
model will be constructed by adopting the hypothesis that the model predicted
using five hidden-layer using 10 historical data as input value is the most
accurate. For weather data, Hadoop, Spark clustering will be configured using
multiple Raspberry Pi to efficiently store and process big data and
semi-structured data. Afterwards, a climate prediction model based on the Keras
Deep Learning model will be implemented using the Elephas external library
[20].
III. METHOD AND DATA
In cities with similar latitudes, the climate tends to be
similar. Therefore, Madrid, Tehran, Seoul and San Francisco, four major cities
with latitude between 37 and 40 degrees, were chosen first. Since then, the
time series of confirmed cases in the city and daily maximum temperature data
observed at observatories in the city from January 1, 1950 to December 31, 2019
have been obtained. San Francisco had a population of about 880,000 and the
other three cities had a population of more than 6 million. It was assumed that
the transmission speed of COVID-19 would decrease after that date when the
maximum temperature of the city was maintained at 16 degrees Celsius or higher
for more than 12 days of the two weeks. Because generally the incubation period
for COVID-19 is two weeks. Therefore, we decided to observe the temperature
change for two weeks. However, there are additional conditions to maintain a
certain temperature for more than 12 out of 14 days, because abnormal weather
conditions may exist during the two weeks observed.
| 