Host intrusion detection with osquery

A critical aspect of maintaining a robust infrastructure security posture is being able to ask low-level question of hosts in your environment. Even on a single host, performing operating system analytics can often be complex, error prone and computationally expensive. This presentation will examine how Facebook is attacking host instrumentation at scale. We’ll discuss how you can use the same software that we use, regardless of your scale or environment, for no cost at all.

Transcript

1. View Slide

2. mike arpaia / facebook
   host intrusion detection with osquery
   

   View Slide

3. javier marcos / facebook
   ted reed / facebook
   mimeframe / facebook
   

   View Slide

4. what’s the
   problem?
   

   View Slide

5. it’s a hard problem
   we’re all trying to catch attackers
   •insider threats
   •espionage

   •external threats
   •APT
   •hacktivists
   •mass malware
   •the list is endless
   

   View Slide

6. we need help
   we’re all deploying tools
   •many of us have too many vendor products
   •often times, a product solves too narrow of a use-case
   •new use-case? new vendor
   •mo’ money and mo’ problems
   

   View Slide

7. we live in a windows centric world
   •more OS X laptops
   •most production infrastructure runs on Linux

   •few are successfully instrumenting their OS X and Linux hosts
   •how would we solve that problem?
   but, times are changing
   

   View Slide

8. desired
   properties
   

   View Slide

9. simple
   performant and reliable
   easy to integrate
   flexible
   

   View Slide

10. simple
    •no complex coding for users
    •low-level details should be abstracted
    •easy to use, deploy and maintain
    

    View Slide

11. performant and reliable
    •host degradation is unacceptable
    •sane resource utilization over time

    •company services should not be impacted

    •extensive logging and metrics
    

    View Slide

12. easy to integrate
    •every company has existing infrastructure
    •distributed configurations
    •real-time logging
    •data warehousing

    •you should have the option to use existing infrastructure to help power
    your host instrumentation
    

    View Slide

13. flexible
    •host instrumentation can help solve many problem domains
    •intrusion detection
    •vulnerability management
    •reliability
    •compliance
    •< insert domain here >

    •having a single solution reduces cognitive overhead and time spent
    

    View Slide

14. osquery
    

    View Slide

15. SQL for your infrastructure
    osquery
    use SQL queries to explore OS state
    •running processes
    •loaded kernel modules
    •active network connections
    •route table
    •firewall settings
    •installed software
    •and more
    

    View Slide

16. osqueryi
    

    View Slide

17. LaunchDaemons which run a binary at boot
    

    View Slide

18. running processes
    

    View Slide

19. processes listening on ports
    

    View Slide

20. more tables are being written every day
    many tables are available
    •alf
    •alf_exceptions
    •alf_explicit_auths
    •alf_services
    •apps
    •ca_certs
    •cpuid
    •etc_hosts
    •groups
    •homebrew_packages
    •interface_addresses
    •interface_details
    •kextstat
    •last
    •launchd
    •listening_ports
    •nvram
    •osx_version
    •passwd_changes
    •processes
    •routes
    •suid_bin
    •time
    •users
    

    View Slide

21. use simple tables, together
    osquery enables complex analysis
    by allowing users to join and
    aggregate across several simple
    tables
    •simple tables have many advantages
    •easier to write
    •easier to maintain
    •can be used in many contexts
    

    View Slide

22. osqueryd
    

    View Slide

23. daemon for low-level host monitoring
    osqueryd
    know how the results of a query change over time
    •schedule a query on your hosts via a config

    •the daemon takes care of periodically executing your queries
    •buffers results to disk and generates a log of state changes
    •logs results for aggregation and analytics
    

    View Slide

24. event-based operating system introspection
    host event pub/sub stream
    subscribe to key OS events to create dynamically growing tables
    •subscribe to “publishers”
    •filesystem changes (inotify, FSEvents)
    •network setting changes (SCNetwork)
    •query the history of your host, as it evolves
    

    View Slide

25. for config distribution and data infrastructure
    plugin system
    •simple plugin API
    •specify your plugins at runtime with a command-line flag
    filesystem
    http
    zookeeper
    configuration
    filesystem
    flume
    scribe
    logging
    

    View Slide

26. #include "osquery/logger/plugin.h"
    #include
    namespace osquery {
    class GlogPlugin : public LoggerPlugin {
    public:
    Status logString(const std::string& message) {
    LOG(INFO) << message;
    return Status(0, "OK");
    }
    };
    REGISTER_LOGGER_PLUGIN("glog", std::make_shared());
    }
    registering a glog plugin
    

    View Slide

27. tables
    

    View Slide

28. creating tables is easy
    easily define what your tables “look like” in Python and use C++ to
    implement what a full-table scan would return
    •the Python is used to generate faster C++ code transparently
    •you write a single C++ function which implements a full-table scan
    

    View Slide

29. View Slide

30. table_name("time")
    schema([
    Column(name="hour", type="int"),
    Column(name="minutes", type="int"),
    Column(name="seconds", type="int"),
    ])
    implementation("time@genTime")
    

    View Slide

31. namespace osquery {
    namespace tables {
    QueryData genTime() {
    QueryData results;
    struct tm* now = localtime(time(0));
    // this could be done in a loop for many rows
    Row r;
    r["hour"] = now->tm_hour;
    r["minutes"] = now->tm_min;
    r["seconds"] = now->tm_sec;
    results.push_back(r);
    return results;
    }
    }
    }
    

    View Slide

32. build and test
    

    View Slide

33. cross-platform build infrastructure
    osquery supports Ubuntu 12.04+, CentOS 6.5 and OS X 10.8+
    •dev tools include vagrant build VMs for all supported platforms
    •every commit publicly builds on Ubuntu and OS X via TravisCI
    •use build VMs to build and package osquery in a way that works for you
    

    View Slide

34. cross-platform build infrastructure
    OS X pkg creation uses homebrew
    to manage dependencies
    •distribute relocatable homebrew
    artifacts such that they work with
    employees existing installation
    $ ./tools/make_osx_package.sh -c ~/Desktop/osquery.conf
    [+] calculating dependency tree
    [+] copying dependencies
    - rocksdb (/usr/local/Cellar/rocksdb/3.5)
    - boost (/usr/local/Cellar/boost/1.56.0)
    - gflags (/usr/local/Cellar/gflags/2.0)
    - glog (/usr/local/Cellar/glog/0.3.3)
    - thrift (/usr/local/Cellar/thrift/0.9.1)
    - lz4 (/usr/local/Cellar/lz4/r116)
    - pkg-config (/usr/local/Cellar/pkg-config/0.28)
    - snappy (/usr/local/Cellar/snappy/1.1.1)
    [+] copying osquery binaries
    [+] copying osquery configurations
    [+] finalizing preinstall and postinstall scripts
    [+] creating package
    [+] package created at ~/git/osquery/osqueryd.pkg
    

    View Slide

35. many tutorials and guides
    extensive documentation
    

    View Slide

36. open source
    

    View Slide

37. all development happens in the open, on GitHub
    work on osquery with us
    the problem that osquery solves isn't unique to facebook
    •https://github.com/facebook/osquery
    •http://osquery.io
    this journey is 1% finished: get involved
    •we’re excited to take on future challenges in the open
    •let’s build together
    

    View Slide

38. what’s next?
    

    View Slide

39. contribute and help us build an awesome tool
    what we’re working on
    there’s a team of great engineers at facebook that are actively
    working on making osquery awesome for everyone
    •more tables
    •ad-hoc remote queries
    •kernel modules for lower-level behavior monitoring
    •deep systems visibility
    

    View Slide

40. questions
    http://osquery.io
    

    View Slide