Caligare - makes the network better Contact Us |  Sitemap |  Our customers   Cisco Technology Developer Partner

Detecting Abnormal Network Activitites With Netflow

Table of Content

NetFlow Overview
What is Netflow?
What is Caligare Flow Inspector?
Baseline Analysis
Top sessions
Top data streams
How to find out if I am being attacked?
Finding infected stations in your network
How to find out who attacked my network?
How to find out who attacked me?
Protection and Prevention


There are a lot of ways to monitor and analyze your network traffic and protect it from Internet intrusions. Organizations commonly use a firewall for network protection. Although firewall logs often provide a huge information regarding intrusion attempts, sometimes might be of too much data to sort through when there is a problem you cannot resolve it quickly. Some organizations also use intrusion detection systems (IDS) on border routers to analyze incoming traffic for patterns that indicate specific problems. But firewall or intrusion detection system is used primarily on borders with the Internet, rather than on internal networks. This is one of reason why Cisco's NetFlow came to the rescue.

Netflow Overview

Netflow is a traffic monitoring and analyzing technology developed by Darren Kerr and Barry Bruins at Cisco Systems. Netflow describes the method for a router and/or intelligent switch to export statistics about the dataflow, and this built-in feature is found on most Cisco routers as well as Juniper, Extreme Networks, Riverstone etc. NetFlow technology provides the data necessary to effectively analyze trend and baseline application data as it passes through the network. It can then be exported to a reporting package and can provide the information necessary to manage critical business applications.

What is Netflow?

Netflow is defined as a unidirectional sequence of packets between a given source and destination which means there will be two flows for each connection session, one from the server to client, one from the client to server. In order to distinguish flows from one another, the source and destination addresses, protocol and port numbers are used. The Type of Service and source input interface index are also used to uniquely identify the flow to which a packet belongs. A flow is determined to have ended when it has been idle for a specified length of time, when it has become older than a specified age (30 minutes by default) or when the flow is a TCP connection a FIN or RST has been sent. The router may expire flows more aggressively if it is running out of cache space.

A number of router vendors have implemented their version of netflow, but version 5 is now the most common. For a NDE version 5, every single UDP packet contains one flow header and thirty flow records at maximum. Every flow record is made up of several base fields and the rest which include: next hop address, output interface number, number of packets in the flow, total bytes in the flow, source and destination AS number, source and destination network length and TCP flags (cumulative OR of TCP flags).

What is Caligare Flow Inspector?

Caligare Flow Inspector is a unique network software solution for companies, who need to plan, build, maintain and manage their network and at the same time keep their network more secure and efficient. Caligare Flow Inspector is a web-based bandwidth monitoring tool that uses NetFlow data export to provide detailed traffic statistics that help answer who, what, when, where of bandwidth usage.

CFI software was engineered to create a secure network-monitoring platform based on industry standards that will fit your existing security policies. The results are the ability to monitor in real time, significantly reducing the time it takes to identify problem and troubleshoot. CFI keeps track of what is happening in your corporate network, detecting attacks, and warning you of problematic network users. All information about network activities are archived in a central database.

Baseline Analysis

A baseline analysis is a model describing what "normal" network activity is according to some historical traffic pattern; any other traffic that falls outside the scope of this traffic pattern will be flagged as malicious. A trend analysis reports is the most common and basic method of doing flow-based analysis. In netflow analysis is main focus on records that have some "special high traffic volume" attribute, especially the value of those flow fields that deviate significantly from an established historical baseline. Normally there are two ways to make use of baseline analysis methods: top sessions and top data.

Top sessions

A top sessions means a single host tries to open an abnormally high volume of connections to a single node or block of nodes. The most reasons for these activities are worms, denial of service attacks and network scans.
Common clients connecting to the Internet should keep a relatively normal connection frequency. But if a host is infected with a worm, it will absolutely act different. It will mostly open a huge number of connections to the destination for its attempts to infect the next batch of victims. For the same reason, when a lesser-skilled "script kiddies" is scanning a large block of addresses for certain vulnerable services, we will see especially high volume sessions sent out by that single IP address.
We can also use top sessions method to detect many kinds of network abuses, such as checking the flow records for port 25 connection requests sent out by every single host in real time. In a given duration, for any host, if the statistics of port 25 requests are above a 'normal' value, it could be considered to be a spammer or someone infected with some kinds of email worm. It would be better for the Internet as a whole if service providers started using this technology and shut down the spammers upon detection.

Top data streams

A second method of using baseline analysis is top data. This can be defined as a large amount of network data transferred in a certain period of time from a single host to a single destination or block of destinations.
The Top hosts that transfer traffic data to or from the outside in an enterprise should be ranked into relatively fixed groups. If this pattern changes, and a new host suddenly appears in the Top hosts matrix, an alert should be triggered.

How to find out if I am being attacked?

Traffic inspection and analysis is a very complex problem. On the market there are many tools as IDS, network traffic dump or network probes, but lack of them can process big traffic volume (e.g. 10TB/hour). We decided to use netflow data export (NDE) that is widely available on most high-end routers for user tracking and real time data flow analysis. Netflow brings transparent view what is happening in your network. There are several methods how to detect if "your" network is under attack.
  1. Packet size distribution. Many short packets (more than 60%) may signify suspicious traffic.
  2. Many connections from single host to considerable destinations.
  3. Using reserved or private IP address on the Internet.
  4. Excessive number of ICMP messages.
In the latest version of Caligare Flow Inspector software there is implemented packet distribution statistic. In our company we are using small honey pot network (without any real stations) for attack analyzing. Example of attack you can see on the picture 1. On the picture there is almost 100% small packets (more that 70% small packets is typical for network attack in most cases). You can use the following steps to locate the source of the problem and some tips on how to filter suspicious traffic.
packet size distribution
Picture 1: Packet size distribution.

Finding infected stations in your network

NetFlow Inspector software is the ideal tool for detecting worm sources (infected stations) in your network. Trends menu may be used for this type of analysis. The following example gives you information on how to find infected stations in your local network.
classic network hierarchy
Picture 2: Classic network hierarchy.

Log into Caligare Flow Inspector and run the following steps:
  1. Select collector that stores netflow data exports (in our case: router R01).
  2. In the table selector choose current hourly table.
  3. Select statistic: source host distributions.
  4. Set source interface (Gigabit Ethernet 1/1).
  5. Set destination interface (not Gigabit Ethernet 1/1).
  6. Run search query.
After displaying source host distributions you can view top ten source IP addresses sorted by number of used unique destination IP addresses (see picture 3). These source IP addresses are candidates on the infected stations.

Check result and select possible infected stations (infected station pool more than 500 unique destinations in most cases). Ignore your servers that are normally heavy used. Web or application servers normally generate many connections to many destinations.
top source host distribution
Picture 3: Top source host distribution (IP addresses are hidden).

Write top 5 sources to notebook and then continue to infected station confirmation step. For each candidate IP address run the following query:
  1. Set statistic: destination ports by packet.
  2. Source IP address: <test_IP_address_from_notebook>
  3. Run search query.
Check destination ports that are in use by potentially infected station. In most case (when station is infected) you will see some of following ports: netbios (137, 138, 139), microsoft-ds (445), ms-sql-s (1433), www (80, 3128) etc (see picture 4).

Now, is a good time to consider if your candidate is infected or not. Decision is yours, because only you know "your" network and servers. If a station opens more than 500 unique destination connections to port 1433, this seems like very suspicious activity.
top destination ports by packet
Picture 4: Top destination ports by packet.

How to find out who attacked my network?

Picture 5 shows a typical network attack. The infected station tries to open a connection to all the servers in your network. You can simply locate this attack by finding the source host that is trying to open a connection to various destinations in your local network.
typical worm attack by infected stations
Picture 5: Typical worm attack by infected stations on your local network.

Check caption "Finding worm sources in your network" and how to find these source hosts. Sophisticated worm sources do NOT pool your whole network, but instead randomly or pseudo-randomly try to open from time to time a single host connection. Locating these attackers is difficult but NOT impossible! You can use TCP flags and ICMP tracking. When the attacker tries to open the TCP connection to an unused destination IP address the TCP SYN flag is set. If the connection is successful you will see cumulative TCP flags SYN and ACK, if the connection is unsuccessful you will see only flows with SYN flag. You can count the unsuccessful connections for every source IP address outside your network and source, the one with the most of connections found is your attacker candidate. If attacker is using UDP protocol and pools your whole network, an excessive number of ICMP messages will then be generated.

How to find out who attacked me?

If you suspect (or know) that your station is victim to an attack, then you probably want to know who is the attacker. Locating the attacker is simple if source IP address is NOT spoofed. Select "Trends" menu and use "Source host by packet" statistic. Type in your IP address (victim) into destination host field and run search query. Result is a list of source hosts who communicated with you sorted by number of packets. Often the first host located is the attacker. In case source IP address is spoofed (often used reserved or private IP address) you can only locate source interface through that malicious traffic going into your station. You can not filter this attacker if he uses random source IP address, you can only contact provider or your ISP peer operator.

Protection and Prevention

You can use many protection mechanisms, these are widely available through access lists on Cisco routers.
  1. Create new access list: ip access-list extended <list_name>
  2. Add block rule: deny ip <wild mask> any
  3. Repeat step 2 for each attacker
  4. Permit any other traffic
  5. Check access list rules: show ip access-list <list_name>
  6. Apply access list on source interface: ip access-group <list_name> in
configure terminal
ip access-list extended block_attacker
 deny   ip any
 deny   ip any
 deny   ip any
 permit ip any any
 permit pim any any
 permit igmp any any
interface GigabitEthernet 1/1
 ip access-group block_attacker in
Be very careful before updating access list! On many routers the default rule is drop any traffic if access list exists. We recommend removing access list from interface then creating a new access list and reassign it to interface. On picture 3 is the result of applying access list on our router R01 that was applied at 10:03.


This attack detection manual has discussed the flow-based analysis of malicious traffic and abnormal activities. With top sessions and top data methods, network administrators can simply detect network anomalies in real time more effectively. There is no universal process on how to find source of attack, but with Caligare Flow Inspector software we may make your corporate network run better.

Copyright: Caligare Company
Last-Modified: November 4th 2005

  © 2003-2024 Caligare. All Rights Reserved.  Terms of Use  | Privacy Statement  | Site Map  | Contact Us