Helen Gibson, University of Northumbria, UK. helen.gibson@northumbria.ac.uk PRIMARY
Paul Vickers, University of Northumbria, UK.
paul.vickers@northumbria.ac.uk
Student Team: YES
MySQL – the IDS and Firewall logs were imported
into a MySQL database which was accessed by R. TPP (Targeted Projection Pursuit
tool) can also access the database to show detailed information about the
nodes.
R and RStudio were used to process the data into an
aggregate format suitable for use in TPP
Weka (http://www.cs.waikato.ac.nz/ml/weka/) was
used to convert and normalize the csv files produced by R into ARFF format
TPP, originally developed by Joe Faith (http://code.google.com/p/targeted-projection-pursuit/),
a high dimensional data exploration tool [1] already extended for graph layout [2] was further extended and then adapted for use in the challenge.
Background on TPP: TPP is an interactive
high-dimensional data exploration tool that allows users to explore two
dimensional projections of high dimensional data sets. By moving the points a
user is requesting a particular projection of the data. TPP returns the view of
a projection that most closely matches the one requested by the user. How much
each dimension contributes to the overall projection is shown in the right hand
table. This, along with the ability to compare the mean value of each of the
selected nodes with the overall mean for that attribute, allows the user to
relate the projection back to the original data.
Video:
Answers to Mini-Challenge 2 Questions:
MC 2.1 Using your visual analytics tools, can you identify
what noteworthy events took place for the time period covered in the firewall
and IDS logs? Provide screen shots of your visual analytics tools that
highlight the five most noteworthy events of security concern, along with
explanations of each event.
The
IDS and firewall logs were aggregated so that they could be used in TPP. This
means counting the number occurrences of each event for each IP address. For
the IDS logs this was the priority, the classification and the label of the
event giving 29 possible attributes. For the firewall logs attributes were 15
minute intervals and the number of events each IP address was involved in in
that interval became the value of the attribute. Since we are converting
edge-attribute data into node-attribute data we also repeated the aggregation
process with the attributes split into source or destination events. A list of
edges was extracted which did not consider the weight of each edge.
Key to Nodes
Noteworthy Events
The
first set of noteworthy events can be seen in the IDS log for day 0406.
Figure 1: Overview of IDS log from day0406.
There are some outliers. Edges are coloured by source node and curved clockwise
indicating direction (larger
version #1)
Event 1: Unusual attempts access to the database and email ports
Figure 2: The five outlier workstation that connect to the firewall interface to
the regional bank network. The table shows which events are causing them to
appear as outlier workstations (larger
version #2)
There are five workstations that appear as
outliers. These all connect to the firewall interface to the regional bank
network. They are:
·
172.23.231.69
·
172.23.232.4
·
172.23.241.156
·
172.23.234.58
·
172.23.236.8.
·
From the table in Figure 2 it is clear that these nodes are involved in events that none of the
other workstations are. These events relate to attempts to access the database
servers in particular. In Figure 4 we see that in the space of a couple of minutes the IP address
172.23.241.156 makes an SNMP request followed by 4 attempts to access each of
the database ports and then VNC scans on 5800-5820 and 5900-5920. The SNMP
requests have Xrefs to the URLs:
http://cve.mitre.org/cgi-bin/cvename.cgi?name=2002-0013
http://cve.mitre.org/cgi-bin/cvename.cgi?name=2002-0012
http://www.securityfocus.com/bid/4132
http://www.securityfocus.com/bid/4089
http://www.securityfocus.com/bid/4088
These indicate that SNMP could be used to remotely
execute code on a user’s machine or be involved in a denial of service attack.
This pattern is similar for the other IPs.
IP addresses 172.23.231.69, 172.23.236.8
and 172.23.234.58 are also involved in potential SSH scans and outbound SSH
scans which indicate a potential brute force attack. IP address 172.23.231.69
also tries to access the email ports and is involved in many database, SNMP and
VNC events.
Figure 3: Through connecting to the IDS
database we see there is a pattern to the attack (larger
version #3).
Event 2: Unusually fast terminal
server traffic
Figure 4: The workstation with IP address 172.23.231.69
(larger version #4).
IP address 172.23.231.69 also raises an
event related to unusually fast server traffic. In the IDS log we see an
associated URL (http://threatpost.com/en_us/blogs/new-worm-morto-using-rdp-infect-windows-pcs-082811).
This informs that an event of this type could indicate the presence of a worm
in the system. An excessive number of TCP connections on source port 3389 are
also associated with this event.
Event 3: DNS and remote code execution
Figure 5: The IDS logs from both days using
source and destination attributes (larger
version #5)
Figure 6: Zoomed view of the workstations
connected to the DNS. We also see here the only link between two workstations (larger
version #6)
Nodes in this group are connecting to
the DNS. They are involved in four kinds of events:
·
GPL
NETBIOS SMB IPC unicode share access
·
GPL
NETBIOS SMB Session Setup NTMLSSP unicode asn1 overflow attempt
·
GPL
NETBIOS SMB DS IPC unicode share access
·
GPL
NETBIOS SMB DS Session Setup NTMLSSP unicode asn1 overflow attempt
In the data the NTMLSSP events have a
number of associated URLs.
http://www.microsoft.com/technet/security/bulletin/MS04-007.mspx
http://cgi.nessus.org/plugins/dump.php3?id=12065
http://cgi.nessus.org/plugins/dump.php3?id=12052
http://cve.mitre.org/cgi-bin/cvename.cgi?name=2003-0818
http://www.securityfocus.com/bid/9635
http://www.securityfocus.com/bid/9633
Each of these describes the fact that
vulnerability exists that allows for the remote execution of code on the host
machine. This may be how the antivirus software is being installed to run on
users’ computers.
Event 4: Large number of connections for two nodes in the firewall logs
Here we use the number of connections
in each 15 minute period as attributes. This results in a graph as in Figure 7. The two workstations have the IP addresses 172.23.0.132 and
172.23.252.10. Workstation with the address 172.23.10.252 has most connections
between 06/04/2012 17:55 and 07/04/2012 09:10 while 172.23.132 has consistently
high numbers of connections which seem to peak in the hour before workstation
172.23.10.252 starts making connections. IP 172.23.0.132 connects to the
headquarters firewall and to BOM accessible websites in the range 10.32.0.X or
10.32.1.X and are all http connections whilst the IP address 172.23.252.10
connects to IPs in the range 10.32.5.X. In this second case many of the
connections are to port 6667 which of often used for IRC which can be exploited
by hackers. There are more than 30,000 instances of port 6667 being used as a
source port and over 2 million as the destination port.
Figure 7: Two IPs have many more connections than other nodes (larger
image #7)
Figure 8: Connections of 172.23.252.10 to BOM accessible websites. The connections to port 6667 all occur within a few seconds. (larger verson #8)
|
|
|
|
|
|
|
|
|
|
|
Event
5: Presence of unknown IP addresses in the system.
In the data preparation each node was
classified according to their IP address. From this emerged a group of nodes
similar to the workstations but have IP range 172.28.29.X. In our
time-attribute firewall graph these nodes have an interesting set of
connections. They both send and receive connections from BOM accessible
websites and the HQ firewall. While some workstation nodes connect to the HQ
and to the BOM accessible websites none of them have connections returned;
although, these incoming connections were actually denied. Nevertheless there
should not be unknown machines connected to the network. These connections
appear for short periods of time: from 05/04/2012 18:22 to 06/04/2012 00:27 and
from 06/04/2012 18:06 to 07/04/2012 00:57; the same time period each night.
Figure 9: The connection of the nodes with
the unknown IPs in the range 172.28.29.X (larger
image #9).
MC
2.2 What security trend is apparent in
the firewall and IDS logs over the course of the two days included here?
Illustrate the identified trend with an informative and innovative
visualization.
The trend in the network seems to be
the numerous opportunities for the potential remote execution of code on the
network. In particular, workstation 172.23.231.69 is involved in many of these
events. In Figure 4 we see that it is involved in an event that may be responsible
for the presence of a worm in the system. In Figure 2 we see it as an outlier to the most of the rest of the
workstations and that it is involved in events that try to access the database
ports, the email ports (Figure 10 below), a possible brute force attack and through the SNMP events
which are another opportunity for remote code execution. In Figure 11 below we also see that when we pinpoint this node in the firewall
log it has also tried to connect to FTP and Telnet ports which is strictly not
allowed.
Figure 10: IP address 172.23.231.69 tries to connect to the email ports (larger
version #10).
Figure 11:IP 172.23.231.69 detected making ftp and telnet requests in the
firewall log (larger
version #11).
As previously mentioned, IRC can be
exploited by hackers. If we look at either of the firewall or IDS logs we see
that most of the workstations and BOM accessible websites are along two axes.
The further along the workstation or BOM accessible website is along these
axes, particularly in the IDS log, the great number of IRC events that have
been detected from these nodes.
MC
2.3 What do you suspect is (are) the
root cause(s) of the events identified in MC 2.1? Understanding that you cannot shut down the
corporate network or disconnect it from the internet, what actions should the
network administrators take to mitigate the root cause problem(s)?
The events seem to indicate that
outsiders have been able to access the network and it has been exploited so
that the installation of this fake antivirus software and scanning program can
be run. It is unlikely that either of these programs is legitimate. The root
causes are that is it too easy for this to happen and it seems to have come
from a number of vulnerabilities in the system. In order to fix this the
following steps should be carried out.
·
All
workstations that exhibit the symptoms of the virus need to cleaned and have
their software reinstalled.
·
The
suggested software patch from Microsoft should be installed onto the network
from here in order to eliminate the vulnerability (http://technet.microsoft.com/en-us/security/bulletin/ms04-007)
·
The
Morto worm is possibly associated with weak passwords and so all workstations
should have their passwords reset and rules about remote desktop connections
should be reviewed as this is how the infection is spread. The worm is possibly
hosted on machine 172.23.231.69 this needs to be immediately
taken offline and the worm removed.
·
The firewall rules should be updated so
that no unknown IP addresses can be on the network.
·
Activity on the IRC channel should be
closely monitored.
References
[1] J. Faith, “Targeted Projection Pursuit for
Interactive Exploration of High- Dimensional Data Sets,” 11th International
Conference Information Visualization (IV ’07), pp. 286-292, Jul.
2007.
[2] H. Gibson and J. Faith, “Node-attribute
Graph Layout for Small-World Networks,” in 2011 15th International
Conference on Information Visualisation, 2011, pp. 482-487.