Week Topic Code Access vs Event Based.pptx

Week 15-16
Code Access vs. Event Based Security

BASIC SECURITY ISSUES:
 Confidentiality
 Disclose information only to authorized users
 Integrity
 Ensure that data is not modified without authorization
 Availability
 Decide who has access to information and how to
make access effective
 Authentication
 Identify a user securely
 Authorization
 Define a set of allowed actions for authorized users
 Non repudiation
 Log users, their actions, and the objects used.
Threats
 Stride model
 Spoofing identity
 Tampering with data
 Repudiation
 Information disclosure
 Denial of service
 Elevation of priviledge

Mitigating Threats
 Protection
 Make successful threats difficult
 Virtually always possible for attacks to succeed in complex
systems if the attacker has enough resources
 Detection
 Detect an attack in progress
 Response
 Put in place plans for each identified threat type
Spoofing
 Attacker claims false identity
 Claim client identity
 Gain access to sensitive data
 Run dangerous processes on server
 Gain administrative priviledges
 Claim server identity
 Collect sensitive data from clients
 Provide false data to clients
 Protection
 Strong authentication using crytography

Tampering with Data
 Tampering with persistent data
 Modify password tables
 Deface web pages
 Add viruses or Trojan horses to files
 Modify routing tables, DNS, …
 Modify audit logs
 Tampering with network packets
 Protection
 Hash codes, digital signatures, encryption
Repudiation
 Attacker denies an action which is hard to
verify
 Claim didn’t delete file
 Claim didn’t make purchase
 Claim didn’t receive services
 Protection
 Audit logs
 Require receipts
 Digital signatures

Information Disclosure
 Attacker sees sensitive data
 Local files
 Network packets
 Attacker gets information about system architecture
 Banners that display software type and version
 Protection
 Strong authentication and access control
 Encrytion of sensitive data
 Turn off banners, disable tracing and debugging
 Use terse error messages
DENIAL OF SERVICE
 Syn flooding
 Send partial connection requests to overflow
server’s list of pending connections
 Distributed Denial of Service
 Amplification attacks
 Send icmp echo requests to broadcast address,
spoofing victim’s address
 Protection
 Install server patches
 Block incoming icmp requests
 Throttle anonymous requests

ELEVATION OF PRIVILEDGE
 Gain administrator priviledge
 Buffer overflow attack
 Overflow designed to overwrite return
address of function with address of malicous
code.
 Exploit bugs in operating system
 Protection
 Avoid buffer overflows
 Validate user input
 Use principle of Least Priviledge
 Grant smallest set of priviledges needed to
function
 Patch operating system
SOCIAL ENGINEERING THREAT
 Social Engineering and Manipulation
 Trick someone into disclosing information, e.g., passwords, credit card ids, SSNs, …
 Manipulation Threats
 Phreaking
 Use fee-based services without payment
 Phishing
 Acquire sensitive information by luring attacks, usually by email.
 Pharming
 Redirecting DNS routes to unauthorized sites to get sensitive information.
 Your on-line banking redirected to Afganistan, for example.
 False representation
 Pretend to be service technician, manager, account owner, …
 Inside Job
 Employee conspires with attacker to compromise system or information.

SECURITY MODELS
 PDR
 Protection
 topic of this presentation
 Detection
 detect attacks
 Response
 plan responses for each threat category
• Call the sysAdmin
• Call the police
• Contact the FBI
• Contact Federal Trade Commission
 AAS - .Net Web Model
 Authentication
 Accept and validate credentials presented by user
 Authorization
 If validated, determine right to access some resource
 Secure communication
 Secure the channel
• SSL, IPsec
 Secure the data
• signing, hashing, cryptography

Windows and .Net
 Role-based
 Authenticate and authorize users, groups,
and accounts (System, Local service,
Network service)
 Actions are authorized through permissions
 Evidence-based or Code Access Security (CAS)
 Code is elevated to the security status of a
user.
 Authorization is based on evidence
• url, zone, publisher, strong name,
custom assembly attributes
 Actions are authorized through policies
WIN SECURITY DEFINITIONS
 Definitions for people and groups of people
 SID – Security IDentifier
 Data structure used to identify user or group.
 Access Token
 A data structure that holds a SID for a security principal, the SIDs for each
group the principal belongs to, and a list of the principal’s priviledges on the
local computer.
 Principal
 An account holder that is assigned a SID for access to resources, e.g., user,
group, service, or computer.
 Combinations of people and objects
 Security Context
 Set of rules for a user’s actions on a protected object
 Combination of user’s access token and object’s security descriptor
 Security Policy
 Rules that define the allowable contexts and mandatory groups.

ROLE-BASED SECURITY
 Use role-based security in programs to control access to methods or
properties at run-time.
 Host authenticates user and provides identity and role information to
CLR.
 Uses NTFS access control lists, IIS security settings.
 CLR makes that information available to code via APIs and permission
demands.
 Can isolate security from code using attributes defined in
System.Security or EnterpriseServices
 System.Security is limited to Windows user groups
 EnterpriseServices uses COM+ roles
 Classes have to inherit from EnterpriseServices
 Which to choose?
 If application has both managed and unmanaged use COM+.
 If application is entirely managed then System.Security is
appropriate.
CODE ACCESS SECURITY
 Goals
 End-user experience
 Managed apps just run
 Safe defaults, no run-time decisions needed
 Administrator
 All settings in one place and easy to customize
 Simple policy model
 Security administration tools
• .Net configuration, CASPOL
 Developer
 Focus on application, security comes free
 Easy to understand and extend when needed

MOBIL CODE
 Old Model
 Obtained from a network, often via a web page.
 Without CAS have either full trust or no trust.
 User decides whether to run.
 If run, code has all the user’s priviledges.
 Inproc COM component, when loaded, becomes
part of the process.
 Can’t distinguish between library code and original
application code.
 CAS model
 Operation based on evidence.
 Allowed actions can be defined at very detailed
level.
 Each assembly can have its own security context.
EVIDENCE-BASED SECURITY
 Definitions
 Permissions
 Objects that represent specific authorized actions
 Permission grant is an authorization for an action given to an
assembly
 Permission demand is a security check for specific grants
 Policy
 Set of permissions granted to an assembly
 Evidence
 Inputs to policy about code
 All three can be extended using security APIs.

STANDARD PERMISSIONS
 Permissions for framework resources
 Data, environment, file IO, Message Queue,
reflection, sockets
 Directory services, event log, web, performance
counters, registry, UI
 DNS, file dialog, isolated storage, printing, security
system
 Identity permissions
 Publisher, site, string name, url, zone
 User identity permission
 Only non-code access permission in Framework.
CODE ACCESS SECURITY IS EVIDENCE-BASED?
 Most permissions are code access
 Demanding permission performs a stack walk checking for grants of all
callers
 Two ways to make checks
 Imperative – call a method
 Declarative
• Attributes in code
• Attributes in configuration file
 Get security by
 Calling class libraries in Framework
 Calling application code with checks

 HOW IT WORKS?
 Loader extracts evidence from assembly
 Evidence is input to policy
 Each level, Enterprise, Machine, User, and
AppDomain, are evaluated
 For each level the union of grants for each
matching code group is determined
 Intersection of permissions from each of
these levels are granted to the assembly
 Apply any assembly permission requests
 Result is the permissions granted to the assembly.
STACK WALK MODIFIERS
 Assertions
 If code vouches for its callers then checks for permissions stop here.
 Gatekeeper classes
 Managed wrappers for unmanaged resources
 Demand permission to call unmanaged
 Assert permission to call unmanaged
 Make the call to unmanaged

 CODE ACCESS CONTROL
 Identity permissions can apply to code as well as users and
groups
 Based on evidence – signature, location, …
 Declarative checks made by JIT at compile-time.
 Imperative checks made by CLR at run-time.
POLICY
 Process of determining what permissions to grant to code.
 Per-assembly basis
 Policy levels
 Enterprise
 Machine
 User
 Application domain
 Each policy level is a collection of code groups
 All code, internet zone, intranet zone, site, strong name (MS Office),
publisher
 Permission grants are intersection of policy levels and union of collection of code
groups.
 Code gets only permissions common to Enterprise, Machine, user,
AppDomain
 Gets all permissions of all groups to which it belongs.

DEFAULT POLICIES
 Local Computer
 Unrestricted
 Intranet
 Limited read environment, UI, isolated storage,
assertion, web access to same site, file read to
same UNC directory
 Internet
 Safe UI, isolated storage, web access to same site
 Restricted
 No access, can’t execute
 Strong name (Framework classes)
 Unrestricted
FRAMEWORK SUPPORT
 Classes used to represent evidence
 Zone, Url, Site, ApplicationDirectory, StrongName, Publisher, Hash
 Classes used to represent permissions
 DBDataPermission, PrintingPermission, SocketPermission,
FileIOPermission, RegistryPermission, …

EVIDENCE
 Evidence is input to policy
 Strong name, publisher identity, location
 Evidence is extensible
 Any object can become evidence
 Only affects permission grants if some code group
condition uses it
 Hosts
 Machine, IIS, ASP.Net, SQL Server
 Fully trusted hosts specify implicitly trusted evidence.
 Semi-trusted hosts cannot provide evidence.
 Hosts can limit policy for AppDomains they create.

Week Topic Code Access vs Event Based.pptx

Recommended

More Related Content

Similar to Week Topic Code Access vs Event Based.pptx (20)

More from ArjayBalberan1 (20)

Recently uploaded (20)

Week Topic Code Access vs Event Based.pptx