This chapter explains the importance of log analysis in today's data-driven world and what are the challenges associated with log analysis. It introduces ELK stack as a complete log analysis solution, and explains what ELK stack is and the role of each of the open source components of the stack, namely, Elasticsearch, Logstash, and Kibana. Also, it briefly explains the key features of each of the components and describes the installation and configuration steps for them.
Logs provide us with necessary information on how our system is behaving. However, the content and format of the logs varies among different services or say, among different components of the same system. For example, a scanner may log error messages related to communication with other devices; on the other hand, a web server logs information on all incoming requests, outgoing responses, time taken for a response, and so on. Similarly, application logs for an e-commerce website will log business-specific logs.
As the logs vary by their content, so will their uses. For example, the logs from a scanner may be used for troubleshooting or for a simple status check or reporting while the web server log is used to analyze traffic patterns across multiple products. Analysis of logs from an e-commerce site can help figure out whether packages from a specific location are returned repeatedly and the probable reasons for the same.
The following are some common use cases where log analysis is helpful:
Debugging is one of the most common reasons to enable logging within your application. The simplest and most frequent use for a debug log is to grep for a specific error message or event occurrence. If a system administrator believes that a program crashed because of a network failure, then he or she will try to find a connection dropped
message or a similar message in the server logs to analyze what caused the issue. Once the bug or the issue is identified, log analysis solutions help capture application information and snapshots of that particular time can be easily passed across development teams to analyze it further.
Log analysis helps optimize or debug system performance and give essential inputs around bottlenecks in the system. Understanding a system's performance is often about understanding resource usage in the system. Logs can help analyze individual resource usage in the system, behavior of multiple threads in the application, potential deadlock conditions, and so on. Logs also carry with them timestamp information, which is essential to analyze how the system is behaving over time. For instance, a web server log can help know how individual services are performing based on response times, HTTP response codes, and so on.
Logs play a vital role in managing the application security for any organization. They are particularly helpful to detect security breaches, application misuse, malicious attacks, and so on. When users interact with the system, it generates log events, which can help track user behavior, identify suspicious activities, and raise alarms or security incidents for breaches.
The intrusion detection process involves session reconstruction from the logs itself. For example, ssh
login events in the system can be used to identify any breaches on the machines.
Predictive analysis is one of the hot trends of recent times. Logs and events data can be used for very accurate predictive analysis. Predictive analysis models help in identifying potential customers, resource planning, inventory management and optimization, workload efficiency, and efficient resource scheduling. It also helps guide the marketing strategy, user-segment targeting, ad-placement strategy, and so on.
When it comes to IoT devices (devices or machines that interact with each other without any human intervention), it is vital that the system is monitored and managed to keep downtime to a minimum and resolve any important bugs or issues swiftly. Since these devices should be able to work with little human intervention and may exist on a large geographical scale, log data is expected to play a crucial role in understanding system behavior and reducing downtime.
The current log analysis process mostly involves checking logs at multiple servers that are written by different components and systems across your application. This has various problems, which makes it a time-consuming and tedious job. Let's look at some of the common problem scenarios:
Non-consistent log format
Decentralized logs
Expert knowledge requirement
Every application and device logs in its own special way, so each format needs its own expert. Also, it is difficult to search across because of different formats.
Let's take a look at some of the common log formats. An interesting thing to observe will be the way different logs represent different timestamp formats, different ways to represent INFO
, ERROR
, and so on, and the order of these components with logs. It's difficult to figure out just by seeing logs what is present at what location. This is where tools such as Logstash help.
A typical tomcat server startup log entry will look like this:
May 24, 2015 3:56:26 PM org.apache.catalina.startup.HostConfig deployWAR INFO: Deployment of web application archive \soft\apache-tomcat-7.0.62\webapps\sample.war has finished in 253 ms
Not only log formats, but timestamp formats are also different among different types of applications, different types of events generated across multiple devices, and so on. Different types of time formats across different components of your system also make it difficult to correlate events occurring across multiple systems at the same time:
142920788
Oct 12 23:21:45
[5/May/2015:08:09:10 +0000]
Tue 01-01-2009 6:00
2015-05-30 T 05:45 UTC
Sat Jul 23 02:16:57 2014
07:38, 11 December 2012 (UTC)
Logs are mostly spread across all the applications that may be across different servers and different components. The complexity of log analysis increases with multiple components logging at multiple locations. For one or two servers' setup, finding out some information from logs involves running cat
or tail
commands or piping these results to grep
command. But what if you have 10
, 20
, or say, 100
servers? These kinds of searches are mostly not scalable for a huge cluster of machines and need a centralized log management and an analysis solution.
The ELK platform is a complete log analytics solution, built on a combination of three open source tools—Elasticsearch, Logstash, and Kibana. It tries to address all the problems and challenges that we saw in the previous section. ELK utilizes the open source stack of Elasticsearch for deep search and data analytics; Logstash for centralized logging management, which includes shipping and forwarding the logs from multiple servers, log enrichment, and parsing; and finally, Kibana for powerful and beautiful data visualizations. ELK stack is currently maintained and actively supported by the company called Elastic (formerly, Elasticsearch).
Let's look at a brief overview of each of these systems:
Elasticsearch
Logstash
Kibana
Elasticsearch is a distributed open source search engine based on Apache Lucene, and released under an Apache 2.0 license (which means that it can be downloaded, used, and modified free of charge). It provides horizontal scalability, reliability, and multitenant capability for real-time search. Elasticsearch features are available through JSON over a RESTful API. The searching capabilities are backed by a schema-less Apache Lucene Engine, which allows it to dynamically index data without knowing the structure beforehand. Elasticsearch is able to achieve fast search responses because it uses indexing to search over the texts.
Elasticsearch is used by many big companies, such as GitHub, SoundCloud, FourSquare, Netflix, and many others. Some of the use cases are as follows:
Wikipedia: This uses Elasticsearch to provide a full text search, and provide functionalities, such as search-as-you-type, and did-you-mean suggestions.
The Guardian: This uses Elasticsearch to process 40 million documents per day, provide real-time analytics of site-traffic across the organization, and help understand audience engagement better.
StumbleUpon: This uses Elasticsearch to power intelligent searches across its platform and provide great recommendations to millions of customers.
SoundCloud: This uses Elasticsearch to provide real-time search capabilities for millions of users across geographies.
GitHub: This uses Elasticsearch to index over 8 million code repositories, and index multiple events across the platform, hence providing real-time search capabilities across it.
Some of the key features of Elasticsearch are:
It is an open source distributed, scalable, and highly available real-time document store
It provides real-time search and analysis capabilities
It provides a sophisticated RESTful API to play around with lookup, and various features, such as multilingual search, geolocation, autocomplete, contextual did-you-mean suggestions, and result snippets
It can be scaled horizontally easily and provides easy integrations with cloud-based infrastructures, such as AWS and others
Logstash is a data pipeline that helps collect, parse, and analyze a large variety of structured and unstructured data and events generated across various systems. It provides plugins to connect to various types of input sources and platforms, and is designed to efficiently process logs, events, and unstructured data sources for distribution into a variety of outputs with the use of its output plugins, namely file, stdout
(as output on console running Logstash), or Elasticsearch.
It has the following key features:
Centralized data processing: Logstash helps build a data pipeline that can centralize data processing. With the use of a variety of plugins for input and output, it can convert a lot of different input sources to a single common format.
Support for custom log formats: Logs written by different applications often have particular formats specific to the application. Logstash helps parse and process custom formats on a large scale. It provides support to write your own filters for tokenization and also provides ready-to-use filters.
Plugin development: Custom plugins can be developed and published, and there is a large variety of custom developed plugins already available.
Kibana is an open source Apache 2.0 licensed data visualization platform that helps in visualizing any kind of structured and unstructured data stored in Elasticsearch indexes. Kibana is entirely written in HTML and JavaScript. It uses the powerful search and indexing capabilities of Elasticsearch exposed through its RESTful API to display powerful graphics for the end users. From basic business intelligence to real-time debugging, Kibana plays its role through exposing data through beautiful histograms, geomaps, pie charts, graphs, tables, and so on.
Kibana makes it easy to understand large volumes of data. Its simple browser-based interface enables you to quickly create and share dynamic dashboards that display changes to Elasticsearch queries in real time.
Some of the key features of Kibana are as follows:
It provides flexible analytics and a visualization platform for business intelligence.
It provides real-time analysis, summarization, charting, and debugging capabilities.
It provides an intuitive and user friendly interface, which is highly customizable through some drag and drop features and alignments as and when needed.
It allows saving the dashboard, and managing more than one dashboard. Dashboards can be easily shared and embedded within different systems.
It allows sharing snapshots of logs that you have already searched through, and isolates multiple problem transactions.
A typical ELK stack data pipeline looks something like this:

In a typical ELK Stack data pipeline, logs from multiple application servers are shipped through Logstash shipper to a centralized Logstash indexer. The Logstash indexer will output data to an Elasticsearch cluster, which will be queried by Kibana to display great visualizations and build dashboards over the log data.
A Java runtime is required to run ELK Stack. The latest version of Java is recommended for the installation. At the time of writing this book, the minimum requirement is Java 7. You can use the official Oracle distribution, or an open source distribution, such as OpenJDK.
You can verify the Java installation by running the following command in your shell:
> java -version java version "1.8.0_40" Java(TM) SE Runtime Environment (build 1.8.0_40-b26) Java HotSpot(TM) 64-Bit Server VM (build 25.40-b25, mixed mode)
If you have verified the Java installation in your system, we can proceed with the ELK installation.
When installing Elasticsearch during production, you can use the method described below, or the Debian or RPM packages provided on the download page.
Tip
You can download the latest version of Elasticsearch from https://www.elastic.co/downloads/elasticsearch.
curl –O https://download.elastic.co/elasticsearch/elasticsearch/elasticsearch-1.5.2.tar.gz
Note
If you don't have cURL, you can use the following command to install it:
sudo apt-get install curl
Then, unpack the zip file on your local filesystem:
tar -zxvf elasticsearch-1.5.2.tar.gz
And then, go to the installation directory:
cd elasticsearch-1.5.2
Note
Elastic, the company behind Elasticsearch, recently launched Elasticsearch 2.0 with some new aggregations, better compression options, simplified query DSL by merging query and filter concepts, and improved performance.
More details can be found in the official documentation:
https://www.elastic.co/guide/en/elasticsearch/reference/current/index.html.
In order to run Elasticsearch, execute the following command:
$ bin/elasticsearch
Add the -d
flag to run it in the background as a daemon process.
We can test it by running the following command in another terminal window:
curl 'http://localhost:9200/?pretty'
This shows you an output similar to this:
{ "status" : 200, "name" : "Master", "cluster_name" : "elasticsearch", "version" : { "number" : "1.5.2", "build_hash" : "c88f77ffc81301dfa9dfd81ca2232f09588bd512", "build_timestamp" : "2015-05-13T13:05:36Z", "build_snapshot" : false, "lucene_version" : "4.10.3" }, "tagline" : "You Know, for Search" }
We can shut down Elasticsearch through the API as follows:
curl -XPOST 'http://localhost:9200/_shutdown'
Elasticsearch configuration files are under the config
folder in the Elasticsearch installation directory. The config
folder has two files, namely elasticsearch.yml
and logging.yml
. The former will be used to specify configuration properties of different Elasticsearch modules, such as network address, paths, and so on, while the latter will specify logging-related configurations.
The configuration file is in the YAML format and the following sections are some of the parameters that can be configured.
To specify the address where all network-based modules will bind and publish to:
network : host : 127.0.0.1
To specify paths for data and log files:
path: logs: /var/log/elasticsearch data: /var/data/elasticsearch
Elasticsearch has a variety of plugins that ease the task of managing indexes, cluster, and so on. Some of the mostly used ones are the Kopf plugin, Marvel, Sense, Shield, and so on, which will be covered in the subsequent chapters. Let's take a look at the Kopf plugin here.
Kopf is a simple web administration tool for Elasticsearch that is written in JavaScript, AngularJS, jQuery and Twitter bootstrap. It offers an easy way of performing common tasks on an Elasticsearch cluster. Not every single API is covered by this plugin, but it does offer a REST client, which allows you to explore the full potential of the Elasticsearch API.
In order to install the elasticsearch-kopf
plugin, execute the following command from the Elasticsearch installation directory:
bin/plugin -install lmenezes/elasticsearch-kopf
Now, go to this address to see the interface: http://localhost:9200/_plugin/kopf/
.
You can see a page similar to this, which shows Elasticsearch nodes, shards, a number of documents, size, and also enables querying the documents indexed.

Elasticsearch Kopf UI
First, download the latest Logstash TAR file from the download page.
Tip
Check for the latest Logstash release version at https://www.elastic.co/downloads/logstash.
curl –O http://download.elastic.co/logstash/logstash/logstash-1.5.0.tar.gz
Then, unpack the GZIP file on your local filesystem:
tar -zxvf logstash-1.5.0.tar.gz
Now, you can run Logstash with a basic configuration.
Run Logstash using -e
flag, followed by the configuration of standard input and output:
cd logstash-1.5.0 bin/logstash -e 'input { stdin { } } output { stdout {} }'
Now, when we type something in the command prompt, we will see its output in Logstash as follows:
hello logstash 2015-05-15T03:34:30.111Z 0.0.0.0 hello logstash
Here, we are running Logstash with the
stdin
input and the stdout
output as this configuration prints whatever you type in a structured format as the output. The -e
flag allows you to quickly test the configuration from the command line.
Now, let's try the codec
setting for output for a pretty formatted output. Exit from the running Logstash by issuing a Ctrl + C command, and then we need to restart Logstash with the following command:
bin/logstash -e 'input { stdin { } } output { stdout { codec => rubydebug } }'
Now, enter some more test input:
Hello PacktPub { "message" => " Hello PacktPub", "@timestamp" => "2015-05-20T23:48:05.335Z", "@version" => "1", "host" => "packtpub" }
The output that you see is the most common output that we generally see from Logstash:
"message"
includes the complete input message or the event line"@timestamp"
will include the timestamp of the time when the event was indexed; or if date filter is used, this value can also use one of the fields in the message to get a timestamp specific to the event"host"
will generally represent the machine where this event was generated
Logstash can be easily configured to read from a log file as input.
For example, to read Apache logs from a file and output to a standard output console, the following configuration will be helpful:
input { file { type => "apache" path => "/user/packtpub/intro-to-elk/elk.log" } } output { stdout { codec => rubydebug } }
Logstash can be configured to output all inputs to an Elasticsearch instance. This is the most common scenario in an ELK platform:
bin/logstash -e 'input { stdin { } } output { elasticsearch { host = localhost } }'
Then type 'you know, for logs
You will be able to see indexes in Elasticsearch through http://localhost:9200/_search
.
Logstash configuration files are in the JSON format. A Logstash config file has a separate section for each type of plugin that you want to add to the event processing pipeline. For example:
# This is a comment. You should use comments to describe # parts of your configuration. input { ... } filter { ... } output { ... }
Each section contains the configuration options for one or more plugins. If you specify multiple filters, they are applied in the order of their appearance in the configuration file.
When you run logstash
, you use the -flag
to read configurations from a configuration file or even from a folder containing multiple configuration files for each type of plugin—input, filter, and output:
bin/logstash –f ../conf/logstash.conf
Note
If you want to test your configurations for syntax errors before running them, you can simply check with the following command:
bin/logstash –configtest ../conf/logstash.conf
This command just checks the configuration without running logstash
.
Logstash runs on JVM and consumes a hefty amount of resources to do so. Logstash, at times, has significant memory consumption. Obviously, this could be a great challenge when you want to send logs from a small machine without harming application performance.
In order to save resources, you can use the Logstash forwarder (previously known as Lumberjack). The forwarder uses Lumberjack's protocol, enabling you to securely ship compressed logs, thus reducing resource consumption and bandwidth. The sole input is file/s, while the output can be directed to multiple destinations.
Other options do exist as well, to send logs. You can use rsyslog
on Linux machines, and there are other agents for Windows machines, such as nxlog
and syslog-ng
. There is another lightweight tool to ship logs called Log-Courier
(https://github.com/driskell/log-courier), which is an enhanced fork of the Logstash forwarder with some improvements.
Download the latest Logstash forwarder release from the download page.
Tip
Check for the latest Logstash forwarder release version at https://www.elastic.co/downloads/logstash.
Prepare a configuration file that contains input plugin details and ssl certificate details to establish a secure communication between your forwarder and indexer servers, and run it using the following command:
Logstash forwarder -config Logstash forwarder.conf
And in Logstash, we can use the Lumberjack plugin to get data from the forwarder:
input { lumberjack { # The port to listen on port => 12345 # The paths to your ssl cert and key ssl_certificate => "path/to/ssl.crt" ssl_key => "path/to/ssl.key" # Set the type of log. type => "log type" }
Some of the most popular Logstash plugins are:
Input plugin
Filters plugin
Output plugin
Some of the most popular Logstash filter plugins are as follows:
date: This is used to parse date fields from incoming events, and use that as Logstash timestamp fields, which can be later used for analytics
drop: This drops everything from incoming events that matches the filter condition
grok: This is the most powerful filter to parse unstructured data from logs or events to a structured format
multiline: This helps parse multiple lines from a single source as one Logstash event
dns: This filter will resolve an IP address from any fields specified
mutate: This helps rename, remove, modify, and replace fields in events
geoip: This adds geographic information based on IP addresses that are retrieved from
Maxmind
database
Some of the most popular Logstash output plugins are as follows:
Before we can install and run Kibana, it has certain prerequisites:
Elasticsearch should be installed, and its HTTP service should be running on port
9200
(default).Kibana must be configured to use the host and port on which Elasticsearch is running (check out the following Configuring Kibana section).
Download the latest Kibana release from the download page.
Tip
Check for the latest Kibana release version at https://www.elastic.co/downloads/kibana.
curl –O https://download.elastic.co/kibana/kibana/kibana-4.0.2-linux-x64.tar.gz
Then, unpack kibana-4.0.2-linux-x64.tar.gz
on your local file system and create a soft link to use a short name.
tar -zxvf kibana-4.0.2-linux-x64.tar.gz ln -s kibana-4.0.2-linux-x64 kibana
Then, you can explore the kibana
folder:
cd kibana
The Kibana configuration file is present in the config
folder inside the kibana
installation:
config/kibana.yml
Following are some of the important configurations for Kibana.
This controls which port to use.
port: 5601.
Property to set the host to bind the server is:
host: "localhost".
Set the elasticsearch_url
to point at your Elasticsearch instance, which is localhost
by default.
elasticsearch_url: http://localhost:9200
Start Kibana manually by issuing the following command:
bin/kibana
You can verify the running Kibana instance on port 5601
by placing the following URL in the browser:
http://localhost:5601
This should fire up the Kibana UI for you.

Kibana UI
Note
We need to specify Index name or pattern that has to be used to show data indexed in Elasticsearch. By default, Kibana assumes the default index as logstash-*
as it is assuming that data is being fed to Elasticsearch through Logstash. If you have changed the name of the index in Logstash output plugin configuration, then we need to change that accordingly.
Kibana 3 versus Kibana 4
Kibana 4 is a major upgrade over Kibana 3. Kibana 4 offers some advanced tools, which provides more flexibility in visualization and helps us use some of the advanced features of Elasticsearch. Kibana 3 had to be installed on a web server; Kibana 4 is released as a standalone application. Some of the new features in Kibana 4 as compared to Kibana 3 are as follows:
Search results highlighting
Shipping with its own web server and using Node.js on the backend
Advanced aggregation-based analytics features, for example, unique counts, non-date histograms, ranges, and percentiles
As you saw in the preceding screenshot of the Kibana UI, the Kibana interface consists of four main components—Discover, Visualize, Dashboard, and Settings.
The Discover page helps to interactively explore the data matching the selected index pattern. This page allows submitting search queries, filtering the search results, and viewing document data. Also, it gives us the count of matching results and statistics related to a field. If the timestamp field is configured in the indexed data, it will also display, by default, a histogram showing distribution of documents over time.

Kibana Discover Page
The Visualize page is used to create new visualizations based on different data sources—a new interactive search, a saved search, or an existing saved visualization. Kibana 4 allows you to create the following visualizations in a new visualization wizard:
Area chart
Data table
Line chart
Markdown widget
Metric
Pie chart
Tile map
Vertical bar chart
These visualizations can be saved, used individually, or can be used in dashboards.

Kibana Visualize Page
Dashboard is a collection of saved visualizations in different groups. These visualizations can be arranged freely with a drag and drop kind of feature, and can be ordered as per the importance of the data. Dashboards can be easily saved, shared, and loaded at a later point in time.
The Settings page helps configure Elasticsearch indexes that we want to explore and configures various index patterns. Also, this page shows various indexed fields in one index pattern and data types of those fields. It also helps us create scripted fields, which are computed on the fly from the data.
In this chapter, we gathered a basic understanding of ELK stack, and also figured out why we need log analysis, and why ELK stack specifically. We also set up Elasticsearch, Logstash, and Kibana.
In the next chapter, we will look at how we can use our ELK stack installation to quickly build a data pipeline for analysis.