Do you think IIT Guwahati certified course can help you in your career?
Introduction
Do you know what cloud monitoring is? Do you know what happens to data stored in the cloud?
In this article, we will learn about Cloud monitoring. It is a method of observing, reviewing, and managing the workflow in a cloud-based infrastructure. These infrastructures include Amazon Web Services (AWS), Google Cloud, and others. This article will cover the advanced concept of Cloud Monitoring. If you want to learn the basics, don't worry. First, read the article Basics of Metrics, Time Series, and Resources in Cloud Monitoring.
If you already know, then let's dive into the article to gather more knowledge about this topic.
Value types and metric kinds
In this article, we will discuss the metric model. This section of the article will explain Metric Type and the attributes that describe it. In detail, two attributes, value type and metric kind are discussed below.
Value type
Value Type is the measurement data type. The value type of each data point is recorded for each time series. The value type describes how the data is saved for measurements. It is only to include one value at a time, such as boolean values and numeric data:
A boolean, BOOL
A 64-bit integer, INT64
A double-precision float, DOUBLE
A string, STRING
The value for distribution measurements is a collection of values rather than a single number. DISTRIBUTION is the value type for distribution measurements.
A distribution has statistics. It summarizes a group of values. Values can be count, max, mean, and other statistics in a distribution. A histogram is also included in a distribution. It will represent the distribution of values across a set of ranges. Latency metrics capture the data as distributions.
Metric kind
The metric kind is included in each time series for its data points. It tells you the methods by which you can interpret the data relative to each other. There are three types of Cloud Monitoring metrics:
Gauge metric: Value is measured at a specific instant of time.
Delta metric: Value is measured at the last recorded change.
Cumulative metric: Value here is increasing over time.
Percentiles and distribution-valued metrics
In this section, we will understand the histogram and percentiles model. It will be with a Distribution value type for metric data. Ranges of values are defined by a distribution metric. It is called buckets. The records of measured value count will fall into each bucket. Distribution metrics report a histogram for counts in buckets. It doesn't report a single measure value.
These value types are mainly used by services. It is when the single measurements are very large in the count for collection. But statistical information about those measurements is precious and valuable. It can include percentiles or averages.
Notes on terminology
There are various words in cloud monitoring whose usage could be more consistent. Name, kind, type, metric, and resources are used in various contexts and combinations. For example 'metric' word can be used in the following context:
A metric type
A “something” that is to be measured
A time series
A set of time series connected with a metric type
The Metric structure is rooted in a TimeSeries structure
Types and names
In a metric descriptor,
Type is a string. It stores the encoded data about the generation of data by the service.
Name value is a qualified metric-type version. It is prefixed with the string projects/[PROJECT_ID]/metricDescriptors/
Name value is predictable within a project from the value of type. But, they are different from each other. However, these terms are referred to interchangeably in casual discussions.
Metric kind and value type
'Value Type' and 'Metric Kind' are sometimes confused with "metric type." But they are not the same. Let's understand why?
Both these terms describe the collected data. They both are related to the ‘data type’ programming concept.
The individual values are described by the Value Type.
How values are related is described by the Metric Kind.
Retention and latency of metric data
This is the last section of this article. Here you will learn how long your metric data is stored by Cloud Monitoring. You will also learn about the latency between data collection and its visibility to you.
Retention of metric data
Cloud Monitoring collects metric data. Then it stores the data for a while in a time series of several metric types. The length of this time varies depending on the metric type.
Monitoring removes the expired data points when that time has expired. Monitoring deletes a time series when all of the points have passed. Deleted time series are not displayed in Monitoring charts or Monitoring API results.
Latency of metric data
When a new metric data point is written, latency is the amount of time it takes before it is visible in monitoring. The time varies depending on the resource being watched and other factors.
Monitoring can collect a measurement in a minute, but you can retrieve that data in up to 4 minutes(latency). So, 4 minutes can be the time stamp for collection recording. Time stamps can vary with the data.
Custom metrics are an option that helps you to reduce this time stamp. You can use the Monitoring API for writing a new data point in an existing time series. It will help you retrieve the data within a few seconds. But writing into a new time series first data point will make you wait for a few minutes. This is for retrieving the data. There will be an extra time required to write in a non-existent time series. This is because it will initiate the time series creation.
Frequently Asked Questions
Why do you need cloud monitoring?
We need cloud monitoring for cybersecurity. As the number of cyber attacks rises, cloud monitoring services must detect possible breaches, identify security gaps, and secure the network well before an attack.
What is the Time Series Metric?
A time series is a series of sequential data points occurring over a particular time interval. A "metric" here refers to the data that is tracked at each increment of time.
What are the metrics in cloud monitoring?
Cloud metrics are data logs that a cloud infrastructure or application generates. Organizations can use the data to detect and monitor. They can also use them to respond to changes in security, costs, and performance of their cloud environments.
Which are the three basic tools for monitoring?
The categories of monitoring are technical monitoring, functional monitoring, and business process monitoring.
Conclusion
In this article, we have extensively discussed Advanced concepts in Cloud Monitoring. We started by explaining to you the value types and metric kinds. Then we covered the confusion that occurred in terminology usage. Retention and latency of metric data are also discussed in detail.
250+ registered 
