1. Field of the Invention
This invention relates to the field of data processing systems. More particularly, this invention relates to the detection of an outbreak of a computer virus on a data processing system.
2. Description of the Prior Art
It is known to detect computer viruses using signature definitions of those viruses. In this technique, when a new virus is discovered, a characteristic pattern of computer instructions is identified associated with the virus and this characteristic pattern then added to a library of computer virus definitions so that a suspect file can be compared against this library of definitions to determine whether it contains one of the known computer viruses identified within the library.
An alternative approach to virus identification is heuristic testing in which types of activity associated with viruses are searched for and used to identify candidate viruses. As an example, an executable computer file that has at its start a call to a routine located at the end of the executable file is a strong candidate for having been infected with a virus.
A problem with both of the above techniques is that they are only generally capable of detecting known viruses or viruses behaving in known ways. As an example, it is only after a virus has been created, released and infected innocent users, that it will be drawn to the attention of anti-virus system makers who can then add a new virus definition to their library of definitions characteristic of the newly released virus to enable its detection by the library technique. Similarly, a virus with a new type of action, such as Word macro viruses when they were first released, present a problem for heuristic identification since their patterns of activity are not ones that are being tested for until after those new viruses have been created, released and infected innocent users.
It is known from Trend's Scan Mail Exchange Beta Release to provide a system in which a virus outbreak is notified if more than a threshold number of known viruses are detected within a 24 hour period. This technique does not address unknown viruses and types of virus behavior.
It is an object the present invention to provide improved anti-virus systems that are able to provide some defense against new viruses and new types of virus.
Viewed from one aspect the present invention provides a computer program product for detecting an outbreak of a computer virus on a computer apparatus, said computer program product comprising:
The invention uses a new virus detection approach. In particular, the invention detects an outbreak of a computer virus by measuring parameters of the activity of the computer system as a whole over a measurement period. The technique does not seek to detect an individual virus by its fingerprint or characteristic manner of execution, but instead measures non virus specific activity. It has been found that when a computer system is subject to an outbreak of a computer virus, then it exhibits certain measurable characteristics that may be useful in detecting the occurrence of the virus outbreak. More particularly, since the technique of the invention is not looking for an individual virus or pattern of execution of a virus, it is able to more readily detect previously unknown viruses by the effect that they have on the activity of the computer system as a whole. The predetermined threshold levels may be manually set or may automatically vary in dependence upon detected long-term behaviour.
As preferred examples of the measurement parameters that may be used there are proposed:
It will be appreciated by those in the field that the usage patterns of a computer system typically vary considerably with the time of day. In an office environment, there is usually considerably higher computer system usage during working hours than outside working hours. Even within working hours, there are often peaks in activity which can have corresponding effects upon the measurement parameters monitored by the invention, e.g. peaks in activity shortly before lunch and in mid afternoon.
In preferred embodiments of the invention the predetermined threshold levels can be varied in dependence upon the time of day and the day of the week.
More generally, the measurement parameters monitored by the system may be selected by the user and the measurement periods over which activity is measured (using a variety of different statistical measurement techniques) may be user selected to adapt the anti-virus system to the particular computer system upon which it is operating.
The invention also provides a method of detecting an outbreak of a computer virus on a computer apparatus and an apparatus for detecting an outbreak of a computer virus on a computer apparatus.
The above, and other objects, features and advantages of this invention will be apparent from the following detailed description of illustrative embodiments which is to be read in connection with the accompanying drawings.
A remote computer 6 may be a virus source. The virus source 6 may be unwitting in that it is merely passing on an infection it has itself incurred. Alternatively, the virus source 6 could be controlled by the virus originator. A computer virus may be introduced into the local area network 2 from the virus source 6 via an E-mail attachment, a security breach of the firewall server 16, a removable media introduced by a user or in some other way.
The local area network 2 is also coupled to a remote computer 4 that is controlled by the anti-virus system provider. This remote computer 4 includes a library of virus definitions 20 and a repository for suspect files 22. The anti-virus provider will frequently update the virus definitions stored by the remote computer 4 to reflect the latest viruses that have been discovered. Computer files suspected by users of the anti-virus system as being infected with a computer virus can be automatically sent to the suspect files repository 22 by the user computers so that they may be analysed as rapidly as possible by the anti-virus system provider.
In operation, the computer system illustrated in
The controlling computer program that controls the computer is stored in the non-volatile storage 28. The computer program may be recorded on a floppy disk, or a CD for distribution. Alternatively, the computer program may be obtained as a product by downloading via the external network link 30.
The computer program executing on the computer system of
It will be appreciated that the way in which the measurement parameters may be derived could take various different forms. The system could look at a rolling average over the measurement period, a peak value within a measurement period, a simple count of instances within a measurement period or various other measurements suited to the particular parameter concerned.
In the case of the above examples relating to E-mail behaviour on a computer system, existing computer E-mail program products, such as Microsoft Exchange Server (produced by Microsoft Corporation), already provide performance monitoring variables that may be read by other programs to gain information concerning the E-mail activity of the computer system. Similarly, many other measurement parameters are already available within computer systems as provided by operating systems or other computer programs executing on these systems. The computer program seeking to detect computer virus outbreaks can read and use these existing parameters. Alternatively, if desired, the computer program may include routines that themselves derive parameters indicative of the activity of the computer system. Conventional programming techniques may be used to derive these parameters.
Step 48 selects the first test from the list together with its associated threshold value and an indication of whether its normal state is above or below the threshold value.
Step 50 detects the measurement parameter MPn associated with the currently selected test. As previously mentioned, this may be read from another computer program or derived by the anti-virus system itself.
At step 52 the detected measurement parameter MPn is compared with its associated threshold value Thn. If the threshold value is crossed, then processing proceeds to step 54 at which a signal indicating a virus outbreak is generated.
If the threshold value is not crossed, then processing proceeds to step 56 where a test is made to see if the last test has yet been reached. If the last test has not yet been reached, then the next test and threshold are selected at step 58 and processing is returned to step 50. If the last test has been reached, then processing terminates.
It will be appreciated that the process illustrated in the flow-diagram of
When a computer virus outbreak has been detected, then the system of the present invention provides an at least partially automated response to that detected outbreak following a predetermined sequence of actions.
When the virus outbreak is detected, step 1 is immediately executed. If the virus outbreak is detected as persisting despite the execution of step 1 and after the expiry of the escalation time associated with step 1, then processing proceeds to seek confirmation that step 2 should be executed. Assuming such confirmation is received, then step 2 is executed and a determination made after an escalation time associated with step 2 as to whether or not the virus outbreak is still persisting. In this way, a predefined sequence of steps are executed spaced by appropriate escalation times set to allow the respective executed step to take effect in order that it may be determined whether or not the virus outbreak has been overcome. In general, the severity and adverse consequences of the various steps in the predefined sequence to the normal operation of the system upon which they reside increases as you progress through the sequence. Accordingly, it is desirable to check after the associated escalation time associated with each step as to whether or not it has been effective since this may avoid the need to execute a more severe counter-measure that would unnecessarily adversely affect the normal operation of the computer system.
The counter-measures that may be taken in the predefined sequence can vary considerably. As examples, given in an order that has been found to provide an appropriate balance between effectiveness and impact upon normal operation, are as follows:
5. Downloading a latest version of the virus definitions file from a remote site in order to increase the likelihood of success of the counter-measures by using the very latest virus definitions. This latest virus definition library may be downloaded from the computer system 4 of the anti-virus system provider via the internet link 8 in a similar way to the preceding step of uploading a copy of the suspect file.
6. An “on demand” scan of all of the computer files stored on the file server 12, the firewall server 16, the mail server 18 and individual client computers 14 of the local area network 2 may be performed.
Such an on-demand scan represents a considerable processing load and is likely to degrade the performance of the local area network 2 while it is taking place, but this may nevertheless be desirable if the computer virus outbreak has persisted despite the preceding counter-measures.
7. Blocking E-mail attachments that appear in excess of a threshold level or blocking all E-mail attachments. The measurement parameters of the E-mail system or bespoke routines within the anti-virus program may detect if particular files or types of files are associated with E-mails being sent or received upon the computer system that has the virus outbreak. If the number of attachments exceeds a predetermined threshold level, then the anti-virus system may interact with the E-mail systems to block further attachments of that file, that file type or all attachments.
8. A recent common type of virus is one that automatically reads a computer user's E-mail address book and distribution lists and then sends itself to those identified E-mail addresses as a way of propagating itself. A counter-measure effective against such viruses is to automatically hide or render inoperative all users' address books or distribution lists such that they may not be read and used by this type of virus.
9. A drastic step that may be taken against a sever virus outbreak is to shutdown the E-mail server 18 and either re-start it in Administrator only mode or not restart it at all. Such a drastic measure is highly likely to be effective against computer viruses using the E-mail services to propagate themselves, but clearly will have a severe adverse impact upon normal use of the computer system.
At step 60, the latest sequence of steps appropriate to the particular time of day and day of week is read. As with the measurement parameters and threshold levels, the predetermined sequence can be varied in dependence upon the time of day and day of week to more appropriately match the use of the system at these times and the availability of support staff to interact with the systems upon occurrence of a virus outbreak.
At step 62 the first counter-measure step is selected. At step 64 a test is made as to whether user confirmation is required prior to execution of the currently selected step. If user confirmation is required, then this is sought via step 66 before processing proceeds to step 68 at which the selected step is executed. If confirmation is not required, then processing proceeds directly from step 64 to step 68.
Step 70 serves to wait for an escalation time associated with the current step after that step has been executed before a test is made at step 72 to determine whether the virus outbreak is continuing. The test applied at step 72 may comprise running the routine illustrated in FIG. 4.
If step 72 reveals that the outbreak has been stopped, then processing ends. If the outbreak is persisting, then processing proceeds to step 74 at which a determination is made as to whether or not the last step in the predetermined sequence of steps has yet been applied. If the last step has already been applied, then processing terminates. Alternatively, if the last step has not yet been applied, then processing proceeds to step 67 at which the next step is selected prior to returning processing to step 64.
A description of the set-up and user interaction with the computer program described above is given in the following description:
Configuration Wizard
The Configuration wizard User Interface (UI) is based on the approach taken in the Microsoft Outlook Rules wizard. The first dialog contains a list of user-defined events and re-actions (rules). These outbreak rules are listed in an order which determines the priority in which determination of an outbreak will occur. The user follows a set of wizard dialogs specifying the data they require for the outbreak event and action. The information is stored in an .INI file to aid in cross-platform portability.
The initial dialog for the wizard (see
When the user presses new or modify they progress through the outbreak wizard pages starting with the one shown in FIG. 8.
Event Wizard Page
The wizard page functions similar to the Microsoft Outlook rules wizard. The user selects an event type in the event list and an English description appears below. There are a series of underlined words. The user clicks on these as if they were html links in a web browser. Upon clicking, a dialog appears asking them to enter a value (see FIG. 9). Once the value is entered it replaces the placeholder but is also underlined and clickable for the user to change/edit.
There is also a threshold event. This allows a user to set, for example, peak mail throughput thresholds for a period in the i.e. am, midday and pm (see FIG. 10).
Upon specifying the event required for an outbreak. The user can click next to move onto the next wizard page. They are not able to progress until the relevant event values have been filled out (in accordance with the table below).
Upon clicking the Next button, the wizard in
Reaction Type Wizard Page
The user is then able to specify the action they require upon an outbreak being detected. There are two categories of action, Manual and Automatic. Upon selection of Manual, the user can specify to be notified via email, network broadcast or pager. They then fill out any notification details using the html type links (See
Once all the values have been filled out for manual notifications, the user continues to either the finish wizard page or the reactions page depending upon what has been selected. If manual reaction has been selected, the reaction page will only appear if they have selected “Use out of office hours”.
Data Validation for Notification
Automatic Virus Reaction Wizard Page
Upon completing the reaction type, the user is presented with the wizard page shown in FIG. 16. The user then has the ability to add and remove items from the list using the buttons at the bottom of the list. The add button will bring up a dialog displaying a choice of available actions to take as in FIG. 17. The move up and move down buttons allow the user to specify the order in which the reactions are carried out. They are then able to specify a time period in which to escalate to the next item in the list. The user can have one or more reaction types in the list. Escalations work down the list from top to bottom. Upon reaching the last item and the event still firing, then notifications will be sent to an administrator via e-mail. An escalation occurs if the event is still firing after the time period for the current action has been exceeded.
The user is also able to specify an outbreak report created (in a file on the hard disk) upon an outbreak being detected. This provides a history of what has happened during detection and automated reaction (an audit trail). The report option defaults to on.
For any action that is considered extreme, the user will be warned via a message box to ensure that they are aware of the data entered.
Data validation for escalation.
Outbreak Summary Wizard Page
Lastly the summary wizard page shows what has been completed during the earlier pages. When the user clicks finish they will return to the dialog shown in FIG. 7.
Outbreak Wizard Flowchart
See FIG. 19.
Outbreak Service
The outbreak service runs as an NT service and can be stopped and started using the services applet in the control panel. The service runs under the system account and therefore can interact with the desktop. There is an icon added to the task bar tray (see
The popup menu provides the ability to view general statistics (i.e. when outbreak thresholds were exceeded and what actions were taken) and outbreak event specific statistics (see FIG. 22). The outbreak event statistics display enough information to the user to be able to intelligently set the outbreak thresholds for the event.
The outbreak service upon a manual event displays the lock down dialog (see
Outbreak events are also enabled/disabled via the popup menu. A tick is placed along side each event that is currently enabled. The popup menu is also able to spawn the performance monitor along with the relevant perfmon work spaces. This will allow the application performance counters to be loaded and configured in performance monitor.
The configuration of the app can also be started from the popup menu. In this case it will load Microsoft Exchange admin.exe.
Appendix A
Events
Actions
Actions will be Manual or Automatic. Manual will notify the user only and not perform any actions unless “Use out of office hours” is specified and the event is triggered during the specified out of office hours.
Although illustrative embodiments of the invention have been described in detail herein with reference to the accompanying drawings, it is to be understood that the invention is not limited to those precise embodiments, and that various changes and modifications can be effected therein by one skilled in the art without departing from the scope and spirit of the invention as defined by the appended claims.
Number | Name | Date | Kind |
---|---|---|---|
5278901 | Shieh et al. | Jan 1994 | A |
5440723 | Arnold et al. | Aug 1995 | A |
5832208 | Chen et al. | Nov 1998 | A |
5956481 | Walsh et al. | Sep 1999 | A |
6016546 | Kephart et al. | Jan 2000 | A |
6701440 | Kim et al. | Mar 2004 | B1 |
20020105545 | Carter et al. | Aug 2002 | A1 |
20020152399 | Smith | Oct 2002 | A1 |
20020194490 | Halperin et al. | Dec 2002 | A1 |
20030167402 | Stolfo et al. | Sep 2003 | A1 |
20040054498 | Shipp | Mar 2004 | A1 |
20040083372 | Williamson et al. | Apr 2004 | A1 |
20040103159 | Williamson et al. | May 2004 | A1 |
20040111632 | Halperin | Jun 2004 | A1 |