On Implementing Real-time Detection Techniques in Future Network Access Control Engr. Arain, Adnan Ashraf
Mehran University of Engineering & Technology, Pakistan adnanlooking@ieee. Ph.D. (Security Models of Wireless Sensor Networks06/09) M.E. (Communication Systems & Networks-2004) B.E. (Computer Systems Engineering-2003)
ICTTA’08, Damascus, 7th -11th April, 2008
Agenda Introduction of ideal network security/
access control Potential threats to network access control General pitfalls of existing “traditional security techniques” The proposed network security technique Real-time test results (pros & cons) Future integration Summary
2
Introduction of Ideal network security Fantastic 4s Integrity Authenticity Confidentiality Availability
AUTHENTICATION
FIREWALL (Client-Server)
BORDER ROUTER
SPECIAL FILTERS (server)
ANTIVIRUS (Client-Server)
DIRECT REGULAR UPDATES FOR EACH
LOG MAINTENANCE
ID and IP SENSORS
VPN DEVICES
`
NEED BASE SERVICE
ALERTS/ NOTIFICATIONS
Host or Client
IPS (Host/Server)
IDS (Host/Server)
Fig 1. Ideal security perimeter
3
Potential threats to security Bad 4s Interception Intrusion Modification Fabrication
4
Defense-in-depth (existing) Border routers Firewalls IDS IPS VPN devices Software services of DMZs Server-side/ client-side web filters Server-side/ client-side antivirus services
“Castel without gate is not the security solution” 5
Defense-in-depth (..continued) AUTHENTICATION
AVAILABIITY
VPN DEVICES
ALERTS/ NOTIFICATIONS
LOG MAINTENANCE
INTEGRITY
DIRECT REGULAR UPDATES FOR EACH
FIREWALL (Client-Server)
` ANTIVIRUS (Client-Server)
IPS (Host/Server)
SECURITY
ID and IP SENSORS
SPECIAL FILTERS (server)
CONFIDENTIALITY
BORDER ROUTER
IDS (Host/Server)
Fig 2. Optimal requirements from a traditional network
6
Defense-in-depth (exemplified by ‘antivirus application’) • General Pitfall of traditional security Server farm AV-update
AV-update Gigabit Switch
AV-update M
Router
10 0
M
bp s
AV-update
AV-update
` Client
s Mbp
AV-update Switch
Switch
100
10/ 100 Mbps
0 10
s bp
10/
7
synchronization of network components/ services due to individual/ independent solutions • Higher traffic load due to recursive dictation from the core-servers for administrating the network. • Non-modularized component services due to monopolization of backdoors by the manufacturer/ developer of the security devices/ application.
1 Gbps
• Minimum
AV-update
`
`
`
Client
Fig 3. Updates in a traditional network
Proposed objectives Maximizing synchronization among security
component services. Modularizing the solution for potential
threats. Minimizing the unnecessary traffic load
8
Switch with resident virus updates
Proposed security solution Requires HELLOs?
• Firewall, secure routing
schemes, authentication mechanisms, antivirus applications should be logically, one entity. • Each of the component
services should be clearly identified and run separately. Routing, ACL, cryptography, and likely resource consuming services must be performed before the arrival of the next packet •
9
NO
YES
HELLO handshake
AV-HELLO handshake
ACK error?
NO
Frame for Authenticity
YES
Check if retransmit HELLOs?
YES
Request to retransmit
NO
Discard MAC entry & Report
Fig 4. Switch-based updates
Implementation of the proposed solution •
AV-Client server
1 Gbps
Core Device
Flash update
Switch x1
` Clients 1-30
10 / 10 0
Mbp s
Switch x2
1 0/
100
Flash update Mbps
Mbps 100
s 0 Mbp 10/ 10
Switch 1
Distribution device 4. 100 Mbps
10
100 Mbps
Flash update
Client 1-30
Minimal sinkholes/ continuous pings/ hello
Distribution device 2
Flash update
`
•
Distribution device 3 100 Mbps
•
Web server Mail server
Mb ps
AV datafiles
Distribution device 1
Regular updates, alerts, REQ/ACK, and log are accessible to clients at near-by locations.
DHCP Authentication & Access server
Corporate firewall server
10/ 10 0
Switches does not generate/ direct the redundant traffic every time towards the coreserver.
Internet
` Clients 1-30
Switch 30
` Clients 1-30
Fig 5. Proposed solution: Campuswide security perimeter
Implementation of the proposed.. (continue) Internet
• Gives protection against
backdoors and trapdoors.
Corporate firewall server AV datafiles
AV-Client server
1 Gbps
Core Device
Distribution device 1
11
Switch x1
` Clients 1-30
10 / 10 0
Mbp s 100
Client 1-30
Flash update Switch x2
1 0/
`
Flash update Mbps
Mbps 100
0 Mbp
s
detected as compared to traditional network security threat.
10/ 10
• Higher number of threats
Flash update
Distribution device 4. 100 Mbps
Flash update
Switch 1
100 Mbps
Distribution device 2 100 Mbps
500+ nodes in campuswide network.
Distribution device 3
10/ 10 0
• Technique was tested for
Web server Mail server
Mb ps
Feasible in existing network infrastructures. •
DHCP Authentication & Access server
` Clients 1-30
Switch 30
` Clients 1-30
Fig 5. Proposed solution: Campuswide security perimeter (Repeat)
Test results of the proposed network security technique Almost 63% better results
2500
Name of attacker
NUMBER OF THREATS DETECTED
Total Number of attacks
Statistic of attacks (blocked) T.N.P.M
P.N.P.M
DOS malware
50
8
16
UNIX viruses
44
8
8
Script malware
232
35
24
Worms
517
111
324
Backdoors
3851
1477
2130
Trojans
3653
725
1461
Other malware
371
279
269
Other OS malware
137
19
8
Total
8855
30.1%
47.9%
TRADITIONAL
2000
PROPOSED
1500
1000
w ar m al O S
ar e
er O th
O th
er
m al
w
Tr
s or ck do
oj an
s
m s or
Ba
tm al rip
W
e wa r
IX N U
Sc
D
O
S
m al
w ar
e
vi ru
se
s
0
e
500
Fig 6 & 7. Traditional versus Proposed network security techniques: Analysis and quantification of malicious codes in the network (Tabular & Graph format)
12
Future extensions UNDERGRAD PROJECTS To develop APIs for CISCO/ Alcatel switches
to update an antivirus application/ firewall policy by some GUI. To re-write flash memory of network switches
in order to reserve memory dedicated for the storage of updates/source in passive mode.
13
Conclusion We discuss the true definition of “network
security perimeter” or “defense-in-depth”.
We
present a hierarchical model with logical separation between different services to analyze the integration possibility of components services.
We show that how synchronization among
14
network component services/ applications may be achieved among clients in their neighborhood core-servers, thus minimizing the traffic load @ core-servers.
Conclusion (..continued) By emulating the proposal we quantify the
results in a real-time network environment. We compare the results of existing and
proposed security technique and found the later one ‘better’. We identify the possible extension to this
research work.
15
Acknowledgement Department of network operations @
Mehran UET, Pakistan Miss: Marvi Mussadiq (IT specialist,
UoSindh, Pakistan)
16
References
17
Hae-Jin Jeong; Il-Seop Song; et-al; “A Multi-dimension Rule Update in a TCAM-based HighPerformance Network Security System”Advanced Information Networking and Applications, 2006. AINA 2006, 20th International Conference on Volume 2, 18-20 April 2006 Page(s):62 – 66 Al-Shaer, E; “Network Security Policies: Verification, Optimization and Testing” Network Operations and Management Symposium, 2006. NOMS 2006, 10th IEEE/IFIP, 2006 Page(s):584 – 584 Magic quadrant, “Symantec network access control; the key to endpoint security” advertising section report 2006, www.symantec.com/endpoint Hamed, H.; Al-Shaer, E; “Taxonomy of conflicts in network security policies” Communications Magazine, IEEE, Volume 44, Issue 3, March 2006 Page(s):134-141 Salim, R.; Rao, G.S.V.R.K;” Design and Development of Network Intrusion Detection System Detection Scheme on Network Processing Unit” Advanced Communication Technology, 2006. ICACT 2006, the 8th International Conference, Volume 2, 20-22 Feb. 2006 Page(s):1023 – 1025 Adnan A. Arain, Marvie, Manzoor Hashmani, “An analytical revelation for a safer network security perimeter”, 2006 proceedings of Intentional Conference on Information and NetworksICOIN2006 Sendai, Japan, 14-17 January 2006 Schaelicke, L.; Freeland, J.C.; “Characterizing sources and remedies for packet loss in network intrusion detection systems” Workload Characterization Symposium, 2005. Proceedings of the IEEE International 6-8 Oct. 2005 Page(s):188 – 196 Jiang-Neng Yi; Wei-Dong Meng; Wei-Min Ma; Jin-Jun Du; “Assess model of network security based on analytic network process” Machine Learning and Cybernetics, 2005. Proceedings of 2005 International Conference on Volume 1, 18-21 Aug. 2005 Page(s):27-32 Vol. 1 Harrison, J.V.; “Enhancing network security by preventing user-initiated malware execution” Information Technology: Coding and Computing, 2005. ITCC 2005, International Conference on Volume 2, 4-6 April 2005 Page(s):597 - 602 Vol. 2
References Yanhui Guo; Cong Wang; “Autonomous decentralized network security system”
Networking, Sensing and Control, 2005. Proceedings, 2005 IEEE, 19-22 March 2005 Page(s):279-282 Stephen Northcutt, Lenzy, et-al, “Inside network perimeter security: The
definitive guide to firewalls, virtual private networks (VPNs), routers, and intrusion detection systems”, ISBN 0672327376, SAMS; 2nd Edition, March 4, 2005 Brian Monroe, Security Engineer-STILLSECURE, “Demystifying network access
control”, white paper 2005, www.stillsecure.com Djordjevic, I.; Phillips, C.; Dimitrakos, T; “An architecture for dynamic security
perimeters of virtual collaborative networks” Network Operations and Management Symposium, 2004. NOMS 2004, IEEE/IFIP, Volume 1, 19-23 April 2004 Page(s):249 - 262 Vol.1 Sean Convery, “Network security architectures”, ISBN: 158705115X, Cisco Press;
2nd Edition Edition, April 19, 2004 Stephen Northcutt, Judy Novak, “Network intrusion detection”, ISBN
0735712654, SAMS; 3rd Edition, August 27, 2002 Todd Lammle, Sean Odom, Kevin, “CCNP- Routing (study guide)”; SYBEX press
Exam 640-503Fdsf Wendell Odom, “CCNA Intro (self study guide)”, CISCO press; Exam 640-821
18
Glen Kunene, “Perimeter security ain't what it used to be, experts Say” Senior
Editor, DevX, www.devx.com/security/ Article/20472
Thanks Q&A
19