The present invention relates to methods and systems for providing an synchronization of directory data and, in particular, to methods and systems for automatically providing seamless synchronization of directory data between different directory server implementations.
Multiple waves of technology have left most organizations with IT infrastructure that is complex, inflexible, and expensive to run. People, devices and applications are tightly coupled making it difficult to rollout new applications or support new work patterns. IT organizations within corporations have to manage a hundreds if not thousands of applications, based on divergent technologies, and run thousands of PCs and servers, each with its own operating requirements and idiosyncrasies.
Maintaining software on a distributed PC or other access device is expensive and time-consuming. As the number of applications and services provided through the PC, the complexity of the PC configuration increases.
Historically this problem has been addressed by ‘locking down’ the PC to limit the changes that can be made to it. Products have also been introduced to ‘push’ software to physical devices but these approaches depend on there being a small, well-defined number of access devices as well as a relatively infrequent update cycle. Until a few years ago this was difficult but achievable in a well managed environment. However, an explosion in the number and type of access devices (which now encompass PCs, laptops, PDAs and mobile phones) combined with a need for frequent, real-time updates (to protect against viruses, worms and security loopholes) has rendered such an approach unworkable.
In large organizations, the problem of access device diversity is compounded by the fact that end users use applications that run on many different platforms. Some run on the user's PC, some run centrally as terminal applications, thin clients or web services and some run on virtual machine technology. Previously, the infrastructure for supporting and managing these applications was entirely separate.
Embodiments of the present description provide enhanced computer- and network-based methods and systems for determining applications to which a user is entitled and providing seamless and uniform access to these applications, regardless of their type and where they are accessed from. Example embodiments provide an Application Workspace System (“AWS”), which enables users to access remote server-based applications (e.g., thin client applications, terminal server applications, applications on hosted operating systems, etc.) using the same interface that they use to access local applications, without needing to know where the application is being accessed. The AWS includes a set of client and server components that implement the notion of a user's virtual workspace, which is presented to the user either seamlessly integrated into the native user interface (such as Windows Explorer) or via a web portal which presents a virtual desktop.
The AWS automatically determines which applications the user is entitled to use, and then figures out automatically, based upon a variety of parameters, which applications are to be made available to the user (resolved to version, particular package etc.), and whether they are to be installed locally, or accessed remotely. Entitled as used herein refers to more than just a concept of access permissions—it involves evaluating a series of rules that specify how applications have been administered for what users and to run in what environments. The parameters that influence this decision define, for example, the user, the access point (the device from which the workspace is accessed), the delivery method (remote, local, etc.) the user's entitlements, application administration parameters, etc. Applications can be associated with users in a “generic” fashion so that they can be “dynamically bound” to a user's access at various times. Thus, the AWS can separate out the administration of applications to users generically (for example, assigning a word processing application to a user) from the administration of actual application implementations that correspond to different versions, packages delivery methods, etc. In this manner, seamless integration of both local and remote applications is made possible.
In the case of applications that are installed locally, the AWS provides for the customization of the configuration and behavior of such applications via scripts that hook system events such as login, power-down, etc. or disconnect or reconnect on terminal servers. In the case of remote applications that are made available, for example, through a terminal server, extensive application behavior modification is supported using a rules-based engine.
The AWS also provides a generalized message bus that can be used for load balancing and resource discovery.
In addition, the AWS provides a generalized LDAP replication mechanism using XSLT stylesheets that can replicate data between heterogeneous schemas. The use of XSLT for replication is extendible to other directory server mechanisms or other information consuming processes that entail heterogeneous data organizations.
The following description details an example embodiment of an Application Workspace System, called “Workspace” or Propero Workspace. Other implementations of the various principles, methods, techniques, components, protocols, etc can be incorporated to provide other embodiments of an AWS. Thus, the present description means to include all such possible embodiments of an AWS.
The Application Workspace System implements an abstraction referred to as a “virtual workspace” that integrates seamlessly into whatever device environment a user uses to access applications. On a personal computer running Windows, for example, once configured, applications appear as icons, which can be “opened” to start an application running (i.e., to launch an application). The AWS automatically configures the user's environment so that the user need not know the particular application that is being run—from a version/package perspective or from a location perspective—an application may be run locally or remotely.
The AWS abstracts the notion of an application so that the task of associating applications with users is separated from the administrative task of mapping particular implementations of the applications to different access devices (from where the application is accessed and how the application is accessed—e.g., a portable or desktop). For example, the directory services (e.g., LDAP) setup of application implementation mappings is an administrative task that need not involve lay users.
To accomplish this abstraction, applications are modeled as belonging to an “application family” that represents the application in an abstract sense (e.g., “Office”). Thus, for example, all kinds of different versions of Microsoft Office may be abstracted as a single family “Office.” Specific applications that can be thought of as instances of those application families (called application objects) delivered via specific means (e.g., “Microsoft Office 2003—Local”, “OpenOffice for Mac—Local”, “Microsoft Office 200—Remote”). This information is stored in an AWS configuration repository, which is typically an LDAP directory, although equivalent storage repositories, protocols, and directory servers may be used.
Stored within the application family object in the configuration repository are a set of rules (sometimes referred to as “policy”) that describes how an application family is to be resolved to a specific application instance at the point of use. Point of use (also referred to as point of access herein) refers to a description of the environment where from the user is attempting to access a particular application. For example, a mobile device being used while traveling abroad typically will be characterized as a different point of use than the user's home personal computer. As a result, the AWS may desire to give the user access to a different application instance than the user normally uses while at home.
Entitlements comprise the definitions (including all the attributes of) the set of application family objects, application objects (including all relevant sub-objects such as document types and launch items), associated package container objects and package objects, along with tuples from all the users user groups, that a given user may have available to him or her at any time. A set of entitlements for a user can be derived from the AWS configuration repository. A process referred to as entitlements resolution further refines a set of entitlements to the set of application objects (and hence the relevant associated objects from the set listed above) that a user actually has available to him or her at a particular point in time.
Assignment of applications (in the general sense) to users is accomplished by creating links in the AWS data model between users and application families—thus allowing for non-IT literate administrators to manage which applications their users need to be able to use without having to be cognizant of the various versions and delivery mechanisms that may by appropriate at specific locations or from specific access devices (such as via a PC or a PDA).
The abstraction of applications breaks the administration of applications into two areas: user administration, which describes which users are associated with which applications families; and application administration which is typically an IT (information technology) function that provides the rules indicating under which circumstances an application family resolves to a specific application. Note that user administration may administer to any defined group instead of to an individual user using any type of grouping mechanism. Groups may also substitute for other entities in the AWS data model.
Note that, for the purposes of this description, the term “application” is used to mean a set of modules, code, script, text or other represented code and/or programs that execute on either hardware machines or higher level virtual machines or are executed by other programs that overall allows a user to achieve some task using the computer (i.e., “apply the computer” to a task). Also, a “package” is a specific set of modules, code, script, text or other represented code and/or programs that execute on either hardware machines or higher level virtual machines or are executed by other programs that are delivered together in some manner. Thus, an application comprises one or more packages. Within an AWS configuration repository, “application objects” are used to represent applications, and “package objects” are used to represent packages.
Note as well that the term “local” or “remote” with reference to applications indicates where, relative to the user, processing occurs in order for the user to achieve the task using the application. Note that this is distinct from where the user interacts with the application through an input device (e.g., screen, keyboard, mouse etc.) as the use of the input device typically will be on the computer the user is physically interacting with.
When a user attempts to login to the user's environment, and at other times, the AWS attempts to automatically determine the virtual workspace for that user from the point of access. In the case of a computing device such as a desktop computer, this determination process results in presenting the user with a set of icons on the desktop that map to particular applications. As an effect of this process, local applications are properly configured, installed, removed, customized and possibly updated, and remote application access is integrated into the virtual workspace. That way, when a user “clicks” on an icon to invoke the underlying application, the “right” thing just happens, be it local or remote. Which applications are installed or configured is determined based upon a process referred to as “entitlements resolution.”
Entitlements resolution is the decision process that resolves an application family object to an application object so that the corresponding application instance can be made available to a user. This is performed at the point of use, and has various pieces of information made available to it from the environment and from the AWS data model. Such information includes, for example, location, network address, access device type, pre-defined tuples (for example, name-value pairs) that may be stored on the access device, and tuples that are associated with any application object in the AWS data model.
Entitlements resolution typically takes place at a point of entry to the system—in many implementations, this can be one of two places:
Note that many different kinds of remotely accessed applications can be integrated into an AWS. For example, an AWS can integrate and support terminal server applications (which run applications in sessions specific to a client access and which are isolated from other applications), Citrix applications (which provide an extension to the general case of Terminal Server applications), Terminal Emulators for character based protocols, virtual machine environments such as hosted operating environments using, for example, VMWare, on-demand application streaming, and any other form of application delivery. The general distinction that the AWS draws between a “local” and a “remote” application is that a local application is some form of code, script or otherwise that executes locally on the user's computer in a manner native to the local computer's native operating system to perform the functions of the application, whereas a remote application has some form of code, script, module or otherwise executing on a computer other than that which the user is using, and any code in that instance that executes locally on the computer the user is using does so only to provide screen, keyboard and mouse functionality for that remote application.
If the entitlements resolution process results in providing a user with access to remote applications, then the mechanism by which they are actually launched (described in detail below) uses web-based components and systems that are common to both access from a workstation running components of the AWS (e.g., a managed workstation) and from a workstation running a browser with a Java Virtual Machine that is used to navigate to a web-based portal (e.g., an unmanaged workstation). In some embodiments, the behavior of remote applications is automatically modified to behave properly when they are launched at their point of use. Behavior modification of applications is also described further below.
Different implementations of an AWS may present different sets of components and services to effectuate the notion of a virtual workspace. In an example AWS, the Propero Workspace (the “Workspace” or “workSpace”), includes a set of tools, modules, data schemas (data models) etc., which are delivered to customers as a set of integrate-able components and/or services. A typical AWS includes: a manager component; a server component; and an analyzer component.
The manager component comprises a set of modules downloaded to a client machine to allow a user to install and run applications locally (or as a “terminal” to remote applications), which are automatically configured and customized on a per-user basis and for the user's environment. A node manager component of the manager component provides a majority of the AWS support for installation and configuration, event based script customization of applications, and communications with the server component.
The server component comprises the entirety of network (e.g., web)-based components that support access from intelligent workstations and access from browsers (via a portal). These components include the ability to perform entitlements resolution, user administration, application administration, application personalization and behavior modification, etc. and implement the virtual workspace.
The analyzer component comprise an extensive set of tools which are used to assist administrators in determining mappings for applications to application families, exclusions, conflicts, dependencies, which users use which applications, etc. A kernel manager component of the analyzer performs application usage based upon detecting a variety of activities—at levels of actual use—not just invocation—on both intelligent workstations and on terminal servers. The detection occurs by “hooking” various system code to be made aware when certain events are triggered.
Example embodiments described herein provide applications, tools, data structures and other support to implement a Application Workspace System to be used for automatic and transparent configuration and customization of applications and seamless integration of remote and local applications in a virtual workspace. Other embodiments of the described techniques may be used for other purposes. In the following description, numerous specific details are set forth, such as data formats and code sequences, etc., in order to provide a thorough understanding of the described techniques. The embodiments described also can be practiced without some of the specific details described herein, or with other specific details, such as changes with respect to the ordering of the code flow, different code flows, etc. Thus, the scope of the techniques and/or functions described are not limited by the particular order, selection, or decomposition of steps described with reference to any particular routine.
Note that the use of the term “workstation” is a matter of ease of description is not meant to convey a specific arrangement of hardware or software. Essentially, any computing device capable of performing the functions indicated can be substituted where this description uses the term “workstation.” Thus, workstations may be mobile or wired device, embedded systems, PDAs, cellular devices, etc.
Node Manager 410 comprises a set of services, modules, and code for supporting the functions of an AWS to implement a virtual workspace. The node manager provides all the necessary functions for packaging, installation, scripting, management, diagnostics, tracing and any ancillary functions that AWS uses. In particular, the Node Manger 410 comprises event runtime and script object support 411; installation, scripting, monitoring, image comparison, and custom services 412; security support 413, and support for network communications 414, for example with the AWS web services.
Note that in some embodiments, some of the code of the Node Manager 410, because it performs “smart” configuration, can also be used by terminal servers to configure remote applications made available through terminal servers, because many of the same functions are needed to configure remote applications. For example, the Node Manager code may also be used for Windows Terminal Servers and Citrix Servers as a system service. This shared code is mainly a figment of efficient implementation, though, and is not a requirement of an AWS.
The user session(s) instances 420 comprise all of the local support needed to interface to local and remote applications using AWS. It typically includes instances of the Node Manager session support 421 that are specific to the user and the particular session; terminal emulation support 422 for interfacing to remote applications; a JAVA virtual machine 423 (although other languages and virtual machines could be similarly incorporated); a web browser 424; and the configured user applications 425. Other components can of course be present as well.
As mentioned, the components of the AWS are used to implement a virtual workspace, which transparently “delivers” access to the right application for the right user for the right point of access at the right time by abstracting out the notion of an application. To support this abstraction, the application information, mappings to other objects, application packages (which define how an application is configured/installed) etc. are set up in the configuration repository typically by an administrator. Application packages are discussed further below with respect to installation of local applications. Application packages can contain scripts that are automatically triggered upon the detection of certain system events (such as logon). Users can also be assigned to application families through the entitlements set up procedures. Once entitlements are set up, users use the workspace to launch applications—in a manner with which they are familiar—and the “right” thing automatically happens. The node manager on a managed workstation or a browser invokes workspace (server) components to automatically configure the virtual workspace, process launch requests, configure applications, etc. In addition, a kernel manager component runs in certain scenarios, such as on terminal servers, to ensure that applications behave properly when they are run remotely. In overview, the kernel manager configures remote applications (such as terminal server applications) to trap certain events, such as file access requests, to ensure that applications behave properly as they may have been written to operate as single user applications and not share resources appropriately.
More specifically, steps 601-610 present an event handling loop. in step 601, when the code detects a logon event, the node manager resolves the entitlements relevant to that user and access point (step 602), transparently installs or upgrades as necessary mandatory applications (step 603), including rebooting the system if the application install/upgrade requires such, and then presents a portion of the virtual workspace with information to allow the user to further configure applications (step 604).
Details of how an administrative sets up application packages for installation are described further below with respect to
Note that once the applications are installed/updated, the user can launch local applications on the managed computing device using the native operating system mechanism. The AWS does not intervene. However, it is possible for someone (e.g., an administrator) to associate shortcuts with specific AWS events, and thus be able to perform custom script actions before applications are launched and at other times in the system.
As mentioned, a user can launch a remote application from a managed computing device as well. Each application object when configured by the AWS contains launch item objects and document type objects. For launch item objects, these consist of a name (e.g., “Microsoft Word”), an icon, and a location attribute (e.g., “Program Files\Microsoft Office”). The AWS Manager generates a shortcut in the location as specified in the location attribute, with the specified icon and name, but referencing a identifier which is generated to include the address of a component in the AWS Web Services, and encoded with a parameter which is set to the identifying key of this launch item. As a result, when a user clicks on one of these shortcuts, a web browser is launched which sends a request to the AWS Web Services which launches the remote application.
Typically, the access device is only aware of the identification key of the remote application. Note that it is common for these keys to point to sub-entities of the remote application (c.f., with an installed application having more than one “icon” that starts a subcomponent of the application). These identification keys are coded into shortcuts or links on the desktop or, if applicable, the portion of the access device's UI commonly used for application related links.
In addition, the system is configured with document types (also known as “file extensions” or “file associations” which are similarly linked to an identification key of the remote application that can handle that document type. There may be various implementations of a shortcut, link, launch item or dock item that represents an identification key that all achieve a common purpose—namely that of notifying a component on the access device that the user has selected or “click” on that shortcut or link.
When a user clicks on one of these links in order to request, or “launch” the associated application (or by selecting a document of a type configured to be associated with a remote application), the identification key, encoded as a URL, is sent to the system AWS Web Services.
Note that, similar to a managed computing device, when a user activates an icon in a web page rendered by the AWS web server on an unmanaged computing device, the mechanism for launching the application is similar. Most of the steps of
In an AWS such as the workspace, users can view and manage existing sessions with remote applications. In the case of an unmanaged computing device, a user can use various user interface (“UI”) controls/elements provided by the AWS Web Services to view and manage existing sessions (such as disconnect them, reconnect them and to kill executing processes within them if available). In the case of a managed computing device, the AWS Manager interacts with the AWS Web Services to retrieve the session data, and then presents this data using its own UI elements (usually handled by the Node Manager Session Support 421 in
To perform session management on the unmanaged device, typically a session controller component of the AWS Web Services queries appropriate other components for any current sessions that exist for the calling user. The session controller compiles a session list, and encodes each session into a Session ID that is unique across all sessions. This session ID has the form:
SessionManagerID;domain\\username(uid)/ID@serverDN:protocol.
The session controller parsers the SessionManagerID to locate a component responsible for the session then the rest of the text is passed to that component. The domain and username identify the user, the ID identifies the actual session itself on the backend server, the server DN identifies the backend server in LDAP, and the protocol is the physical protocol used to connect to that backend server (as some components may support more than one protocol).
As mentioned previously, the AWS abstracts users from applications using a concept of an Application Family.
In order to make the administration of large numbers of users more convenient, they can be grouped together and managed as a single unit. There are two ways of performing this grouping: user folders and user groups. User folders are structures akin to directory folders within a filing system. Each user is contained in a single user folder that may in turn be nested inside a parent user folder. User folders provide inheritance. In the example shown in
User groups are looser structures that allow users from anywhere in the user folder structure to be grouped together, independent of that folder structure. Like user folders, user groups can be used to assign applications to multiple users at one time. User groups cannot be nested inside one another, nor is there any inheritance between them.
Applications in the AWS separate the business-oriented concept of a particular application (like “Microsoft Word”) from the technical details of the package(s) that are used to deliver it (like “office2003.msi” and “office2003sr2.msi”). This separation allows changes to be made to the way that an application is delivered without having to change the entitlements of users to that application. As shown in
Application families allow entitlement assignments without concern for which version or platform of the application is used. For example, the application family “Microsoft Word” could be linked to separate application instances for the Windows and Macintosh editions. Entitlements processing will ensure that users assigned “Microsoft Word” would get the appropriate edition for their platform.
Application groups allow a group of applications to be bundled together and assigned as a single unit. With the example shown in
Application instances describe specific versions of applications; they are what is ultimately presented to users on the ‘Start Menu’ or elsewhere. Generally, application instances are only directly assigned to a user if the user requires a specific version of an application.
So, while Application Families and Package Containers are the objects that are assigned (both in terms of an Application Class to a user, and a Package Class to an application instance), the AWS then decides based on the runtime environment which application (and package if a local application) is actually required and installed. This installation information is then stored on the workstation and associated with the users profile to allow for future decisions (such as exclusions) to be made. Applications as modeled in the data model described by
The term “Application” or “Package” may be used to refer to the configuration item in the configuration repository representing the application or package or it may refer to the application or package itself, depending upon context. In most cases, this description appends “object” when meaning the configuration item in the configuration repository.
User objects are stored in hierarchy within the configuration repository. In an embodiment that uses an LDAP implementation, the AWS uses standard LDAP schema concepts of Organizational Units (OUs) to support this abstraction.
User Group objects contain sets of DNs (Distinguished Names—the LDAP concept of the primary key of an object) pointing to the users in the OU hierarchy, and DNs to application objects, application family objects and application group objects. They exist as LDAP object classes groupOfUniqueNames and pae-UserGroup, stored under ou=Groups, and contain, for example:
cn (Name)
Display Name
DNs of users who are members of the group
DNs of applications
pae-NameValuePairs—strings associated with this group
Application groups are groups of DNs pointing to applications in the application Directory Information Tree (an LDAP term referring to an area within the LDAP directory), and may well point to another application group. They exist as LDAP object classes groupOfUniqueNames and pae-ApplicationGroup, stored under ou=Groups, and contain, for example:
cn (Name)
Display Name
DNs of applications
pae-NameValuePairs—strings associated with this group
Application objects are chosen based on the policy as held in the application family object. A “policy” is a set of rules that describe which application object a particular application family object resolves to at a particular point in time and at a particular point of use. This policy may be “partially resolved” by the entitlements support of the AWS Web Services (as some parts refer to information that is only known at that point, such as the method of authentication to the AWS), and partially resolved by an AWS Node Manager based on information available locally to the managed workstation (such as packages already installed on the computer).
Policy is stored in the application family object as a goal of entitlements resolution is to resolve an application family object to an application object for that user at that time. While it is possible to assign an application object directly to a user object in the configuration store, this may not be advisable as it may result in the system attempting to do something inappropriate (e.g., installing the wrong application on the wrong operating system), or attempting to provide a remote application when there is no network connectivity, etc.
Examples of policy as expressed for local applications are:
The nature of expressing circumstances is through the use of the name-value tuples (e.g., name-value pairs). These name-value pairs are available in the query language, and can form part of the expressions. In addition, the environment and user name-value pairs are available to the expression in the form of expansion variables (that take the form $(Variable)). Finally, some de-facto variables are available, most notably, the key of the application family object in question, and environmental information, such as whether a network connection is present, the IP address and DNS name of the computer, etc.
Each “rule” of the policy exists in an ordered list. This list is then evaluated, in order, with each successful evaluation (i.e., the policy returning a value) resulting in that value being pushed onto the back of a vector which represents the result of processing the policy for that application family object.
Examples using the above policies are as follows.
Example: If one is already installed, use that. Otherwise, application is not available.
Example: Always use this one.
(wsid=‘cn=1,0,0,1,ou=MyApp’)
Example: If nothing else is installed, use this one, otherwise use the one that is installed.
Example: Use this one for this particular operating system.
A policy has three flavors—LOCAL, REMOTE, ALL/BOTH. The three flavors are because there are two different scenarios when a user can access workspace:
PRIORITY:TYPE:USERS:RULE
where:
Once an application family object has been resolved to an application object, if the application object represents a local application, then package objects must then also be determined and associated with the application because packages are the ultimate item that gets physically installed on the computer to enable the application to function. One reason for the existence of package containers is that, when using Windows Installer Packages (MSIs), it is necessary to remove using the path to the old MSI, and install/upgrade using the path to the new MSI. As a result, the locations of both need to be available at the time of an upgrade, so simply modifying an existing package object to point at a new MSI would not be sufficient. The package containers allow administrators to group together the information needed for such removal.
Packages to be installed locally can be vendor supplied installation packages that do not require any further modification or they can be packages generated by the AWS. An example embodiment of the AWS packaging mechanism splits an application package into three separate parts—SYSTEM (containing shared data that is common across the whole computer system), USER (containing information that is stored for each user), and WSM that contains scripts which can perform configuration tasks when certain events are triggered.
The AWS creates a package using a mechanism of capturing the state of a sample access device (similar in nature to the access device that the package will eventually be installed on). This is not a run-time task, but an administrative task. The application is installed from its source media onto that same access device. The AWS then captures the state of the access device after the install as above has occurred, but applies rules to decide which parts of that install can be classified as “SYSTEM” (i.e., applies to all users and does not require access to any user specific configuration areas on the access device), and which parts can be classified as “USER” (i.e., are user specific, and would need to be added to any user specific configuration areas on an access device before a user was able to properly use that application on that access device). The system contains a default set of rules that may be provided to the Node Manager by the AWS Manager Workbench at the time the package is created, but they can be adjusted if required.
This package can then be adjusted using tools available in the system (the Workbench), and can have scripts added to the package that are triggered by the system when certain events occur (see
Once adjusted, this package can then be placed in the AWS package repository—this process also involves configuration data relevant to the package being added to the AWS's configuration database (usually, a replicated LDAP directory).
As mentioned one of the features offered by the AWS application abstraction mechanism is that associations between the concepts of applications and users are actually between users and application families.
Administrators can associate application family objects with users, groups etc. by using direct manipulation techniques (such as drag and drop).
In one embodiment of the AWS, the association between the user and application family is stored in LDAP as a DN link as an attribute of the user object in LDAP pointing to the Application object in LDAP. Where groups are used, the groups act as a form of intersection between the two objects—that is, they contain a list of DNs to the application objects, and a list of DNs to the users who are members of the group.
Configuration and personalization of applications either on the managed workstation (i.e., local applications) or on application servers that are participating in the system (e.g., Terminal Servers) occurs when specific events are generated on these computers. In one embodiment of the AWS, a Node Manager component of the (see, for example, Node Manager 410 in
In order for customization to occur, an administrator needs to have defined scripts and associated them with the application object of interest. Adding scripts to a package to be triggered upon events can be performed using the AWS Manager Workbench. As described above, each package has three parts (called features), which include the SYSTEM, USER and WSM features. WSM stands for workspace Manager, and is the portion into which scripts are stored. In
Table 1 below includes explanations for example events which can trigger the execution of scripts associated with an application package:
Note that scripts have access to all of the tuples (e.g., name-value pairs) configured for that package in the configuration repository, as well as tuples for all applications already installed on the local computer (so that application level settings can be shared), as well as tuples configured for that user and associated groups. (There are a number of different precedent orders that can be applied when merging the set of tuples to deal with tuples that have the same name). These tuples typically exist as a multi-valued string attribute on each object in the configuration repository.
By nature of being able to manage, configure and supply configuration information via these tuples stored in the AWS configuration repository, the AWS as a whole is able to deliver, manage, configure and personalize applications for users. For example, a script that responds to changes in the network state could change the default document location of a word processor to the user's home directory on a network server if the network is present, and to a local directory on the computer when the network is no longer present.
Of note, the scripts execute in an environment that allows full remote debugging of their execution, even though they are asynchronously triggered.
The AWS Node Manager implements an asynchronous event handler for invoking appropriate application scripts when an event is triggered. Events can be system triggered or defined by the AWS. In one embodiment, events are triggered by four distinct mechanisms:
Each time one of these external events occurs, an external event handler (e.g., a WMI subscriber, a SENS subscriber etc.) calls the Node Manager daemon process (either through an internal API or by invoking WSUTIL) with the name of the event (see SCRIPT_EVENT_XXX definitions), specifies the target queue/service name as the script service (“ScriptManager” with ID “{E8F61878-8DA6-494c-829D-017773AF1F05}”), and builds a parameter block if appropriate containing parameters for the event (e.g., a network connect event may indicate which interface was connected at which speed).
In particular, the script name for a package that is part of an application is stored in the installed database (on the managed computing device), as it was an attribute of the package object that led to the installation of that package. Each script is stored in a sub directory named according to the external event type. For example, if a package had a script name of “TEST”, and defined a script for STARTSCREENSAVER and STOPSCREENSAVER, then the actual files on the disk would look like:
By querying the installed database, the Node Manager can build a list of script names to execute. By examining the name of the event as passed to it, it can identify the directory in which to find the scripts. Thus it is able to execute the correct scripts for the event. If no username is passed in, then the Node Manager by default will execute all the scripts for that event.
In step 2106, the handler obtains the relevant name-value tuples that correspond to the applications and their associated family objects and associates them as appropriate with the scripts. The handler then returns to the beginning of the loop in step 2104. In step 2107, when all applications having scripts triggered by the current event have been handled, then the handler obtains name-value tuples for users and for the user's groups. Finally, in step 2108, the accumulated script objects are executed.
In order to be able to provide personalized customization for the applications, each script needs to have access to the correct tuple set for the user, for that application. Therefore, for each script that is executed, a set of tuples (also referred to as NVPs, or Name Value Pairs) is compiled that the script can access.
The installed database records all the tuples associated with users, application objects (and the associated application family object that resolved to that application object) and package objects that Node Manager has installed. These are stored in the installed database with their owning object, allowing the Node Manager to build different sets of tuples at different times. (These sets are refreshed every time Node Manager reads an entitlements document thus allowing new configuration values that administrators may set in the configuration repository to be used even if the application/package has already been installed.)
Therefore, whenever a script needs to be executed, NM builds a set of tuples from the installed database that is correct for that package. The precedence order of building this set is as follows:
Appendix C, incorporated herein by reference in its entirety, provides an example Visual Basic script that configures GAIM Instant Messaging for each user. This is the script used in
Referring back to
Entitlements resolution is the overall decision process that takes information stored in the AWS configuration repository and decides which applications a user is allowed to use, over which delivery mechanisms, and at the point of use. Thus, entitlements resolution is central to providing a virtual workspace. The AWS resolves the application families that are assigned to a user into the actual applications that are to be used at the point in time they are being requested, whether local or remote.
In one embodiment, entitlements resolution is performed by an AWS Node Manager querying the Entitlements Support and Information Services components for entitlement information and then performing a rules-based set of actions as described above in the discussion regarding policy.
Entitlements resolution is typically performed whenever a new user logs into the system as described with reference to
Specifically, in step 2501, the service determines the designated user, either from the request itself (via optional parameter <UID>) or implicitly from an authentication request. In step 2502, the service determines which user the user ID corresponds to (the designated user) and retrieves the groups the user is a member of from the configuration repository. In step 2503, the service determines the name-value tuples that correspond to the designated user and the user's groups. In step 2504, the service determines which application family objects have been associated with the designated user or the designated user's groups. In step 2505, the service determines all of the possible application objects that could be available to the designated user by looking at the application object associations of each determined application family object. In step 2506, the service determines all of the possible package objects and package object containers that correspond to the determined application objects that are local. In step 2507, the service populates an output report, here an XML “entitlements document” with all of the determined and retrieved information from the configuration repository. In step 2508, the output report is returned to the request, and the processing completes.
Thus, after processing, in the example embodiment, the information stored in the XML entitlements document includes:
Appendix A, incorporated herein by reference in its entirety, is an example entitlements document stored as an XML document.
Specifically, in step 2601, the service determines the designated requester (typically a user ID associated with the web browser interface or the AWS Manager). In step 2602, the service requires the information services component (which interfaces to the configuration repository) to retrieve information specific to the requestor. In step 2603, the service queries the information services component to retrieve a profile based upon a set of current attributes, as described above. In step 2604, the service matches the retrieve requestor information to the retrieved profile to make sure the profile is applicable. At this point, in step 2605, the entitlements support sends a request to the Information Services (e.g., the LDAP support) to obtain an entitlements document (as described with reference to
Note that although many of these service routines are shown in the flow diagrams as sequential steps, it is to be understood that many of these services are written as asynchronous event handlers that process requests and data as they arrive. Equivalent synchronous flows are correspondingly incorporated.
At this stage, the entitlements document retrieved from the AWS Information Services contains a list of application family objects that have been assigned to the user or via group memberships, and the partially resolved policy document from the entitlements support contains a partial resolution of the policies contained in the entitlements document. The next step is for the AWS Node Manager to resolve these entitlements into vectors of possible application objects. The reason that vectors are used, rather than single items, is to provide alternatives if a user does not select the most appropriate application (e.g., the one at the beginning of the vector). If another application is available, then the Node Manager will use it instead. For example, if a local install of an application is not desired, and a remote one is available, then the Node Manager will make available the remote application.
In summary, the routine performs a loop on each application family object found in the entitlements document and evaluates each corresponding policy resolving the results where appropriate with the information returned by the partially resolved policy document. Specifically, in step 2701, the routine processes each application family object starting with the first. In step 2702, the routine determines whether there are more application family objects to process, and, if so, continues in step 2704, else continues in step 2703. In step 2704, the routine determines whether there are more policy statements to process, and, if so, continues in step 2705, else returns to the top of the outer loop to obtain the next application family object to process. In step 2705, the routine evaluates the next policy statement in priority order. In step 2706, if the policy statement is tagged as “LOCAL”, then the routine continues in step 2712. If in 2706, the policy is not tagged as “LOCAL,” then in step 2707, then determines whether the policy statement is tagged “ALL” (or “BOTH” in some versions). If so, then the routine continues in step 2709, otherwise continues in step 2708. In step 2708, the policy statement is tagged “REMOTE,” and the routine looks up the corresponding result in the partially resolved policy document (to make sure there is a value), pushes an indicator to the resulting application object onto the application object vector, and returns to process the next policy statement in step 2704.
In step 2709, the routine has determined that the policy statement is tagged “ALL” or “BOTH.” It evaluates the policy and then in step 2710 compares the resulting value with the value returned by the remote results communicated in the partially resolved policy document. In step 2711, if the results match, then the routine pushes an indicator to the resulting application object onto the application object vector, and returns to process the next policy statement in step 2704.
In step 2712, the has determined that the policy statement is tagged “LOCAL.” It evaluates the policy and then in step 2713 determines whether the evaluation resulted in an application object (it succeeded). If so, in step 2714, the routine pushes an indicator of the resulting application object onto the application object vector associated with the current application family object, and returns to process the next policy statement in step 2704.
Once all of the entitled application family objects have been processed and all of the respective policies, then the routine returns a result vector (step 2703) with application object vectors for each application family object.
Once the results vector is completed, then, per step 2408, the Node Manager adjusts each application object vector to account for dependencies and exclusions.
To perform this determination, the Node Manager maintains what is termed an “installed database” on the managed computing system (commonly stored in the Windows Registry) in which it records the keys of all package objects and application objects installed on the computer. This database is partitioned into the SYSTEM part, which records which actions occurred system wide on the computing system, and the USER part, which records which actions occurred for that user. The exclusion check is made against the SYSTEM part of the database. In other words, if the computing system has had an application installed on it that a potential application object lists as an exclusion, then this potential application object (potential in that it exists in the set of vectors above) is then removed from that vector. The purpose of this exercise is to prevent incompatible applications being installed on the same workstation which may occur if two or more users share a workstation, and different applications are assigned to them in the configuration repository.
In step 2807, the routine checks if the current application object being examined has a dependency on one or more additional application objects. If not, the routine continues back to the beginning of the loop in step 2802 to examine the next object. Otherwise, then in step 2808, each of the dependent application objects is added to the vector (or a new vector added for each and added to the application object vector).
After the results vector is has been adjusted, then, per step 2409, the Node Manager processes each application object to determined whether to queue applications for installation or upgrade. This involves using the installed database on the computing device as mentioned above.
In summary, the Node Manager iterates through all of the package objects referenced by each application object, and checks in the installed database to determine whether some or all parts of packages associated with the possible application objects have already been installed and are correct. The Node Manager then decides whether an application needs to be installed or updated and marks it according.
Specifically, in step 2901, the routine processes each application object in each application object vector starting with the first. In step 2902, the routine determines whether it is done. If there are no more objects to process, then the routine continues in step 2913 to return the application object result vector and then starts a new loop (steps 2914-2919) to process any applications that must be installed and will cause a reboot. If there are more objects to process, then in step 2903, the routine determines whether an “Install_Always” attribute has been set, and if so continues in step 2912 (to require an upgrade and potentially mark as mandatory, including identifying packages that are no longer required), otherwise continues in step 2904. In step 2904, the routine determines whether both the SYSTEM and the USER parts have been installed. This check involves checking that the versions of the package objects match as well as that all of the packages have been installed. This includes identifying packages that are no longer required. If in step 2905 everything matches, then the application object is considered to have been already installed correctly. Then in step 2906, if this application object is at the top of its containing vector, the whole vector is removed from the set in step 2907 and the routines skips to the next application object vector in step 2902. Otherwise, in step 2908, the application object is removed from the vector (it isn't the top object) the remainder of the vector is processed in step 2902.
In step 2905, the check for a match could fail either because no corresponding application has been installed, or some portion didn't match. Therefore, in step 2909, the routine checks to determine whether nothing was installed. If true, then in step 2910 the application object is marked as available for installation (which is used for UI hints in the workspace presentation) and the remainder of the vector is processed in step 2902. If false (the application was previously installed and the packages and values don't match), then the routine continues in step 2911 to determine whether the parts that are present or the packages require an upgrade. If so, the application object is marked for UPGRADE in step 2912, and, if at the beginning (top) of the vector, it is marked also as MANDATORY. The routine then skips to the next application object.
Note that this process of deciding what needs to be done is primarily performed to filter out which application objects need no action—this process is repeated later on when applications and packages are actually installed/upgraded/removed.
In steps 2914-2919, the Node Manager then scans all of the application objects at the top of each vector in the set (step 2914) to see if any have both the mandatory attribute set and also have a reboot attribute set (step 2915). The reboot attribute may be used as a UI indicator to warn the user that this application may require a reboot after installation, but if this attribute is present in combination with the mandatory attribute, then in step 2919 the routines causes these applications to be installed and the system rebooted before anything else happens. If the shell (as in the operating system user interface such as Windows Explorer) has not been started yet, it is not started at all if the Node Manage is configured with the “DontShowShellIfRebootRequired” flag is set. This prevents user activity as the machine will be rebooted after these mandatory operations have completed. This approach is commonly used for critical application or security fixes. After the machine reboots, the whole entitlements resolution process is restarted as the user logs on.
In step 2916 (by the time the Node Manager arrives at this step again) then scans all of the application objects at the top of each vector in the set to see if any have the mandatory attribute set. Those that do are queued for installation processing immediately. Immediate processing can be achieved as the Node Manager is an asynchronous queue based system. In step 2917, the Node Manager then scans all of the application objects at the top of each vector in the set to see if any are remote application objects. Any found are queued for remote application processing immediately. In step 2918, the resulting application object vectors are returned for further processing.
In step 2410 of
The Node Manager then continues processing the user selections in step 2411 of
The routine then performs a loop to queue for installation/upgrade all requested objects, or to install/upgrade alternatives if they are available applications (to which the user is entitled).
Specifically, in step 3103, the Node Manager searches each application object vector in the set, starting with the first. In step 3104, if there are no more vectors to process then the routine is done; otherwise, it continues to examine the vector in step 3105. In step 3105, if top application object has been marked as requested, then in step 3106 this application object is submitted for installation processing and the next vector is processed. On the other hand, if the top application object has not been marked as requested, then in step 3107, the Node Manager moves down the vector to see if any further application objects are available. Note that no further choices are presented to the user—this technique is usually used to provide the user with a remote application if the user chooses not to install a local application, or to ensure that an existing local application is upgraded, if required, if the user chooses not to install a different local application. In step 3108, If the NM finds an alternate application object in the vector, and this application object is a local one marked requiring an upgrade, it is submitted for installation processing in step 3109. In step 3110, f the application object is remote, then it is submitted for remote application object processing in step 3111. The routine then continues back to step 3104 to process the next vector.
At this point, the Node Manager will have resolved all the contents of the entitlements document, and will have submitted the relevant application objects for either installation processing, or for remote application object processing (note that “installation processing” performs any steps involved in managing local installs, including upgrades or removals). The Node Manager now deletes the set of vectors, as important state is now held in three places:
The queue of work for installation processing
The queue of work for remote application object processing
The installed database on the managed computing system.
As each application object is processed by installation processing, the installed database is updated. Remote application object processing does not update the installed database, as nothing is actually installed in the database for remote applications.
The Node Manager then triggers the SCRIPT_EVENT_CONFIG script for each application that is present in the installed database as installed for that user.
As mentioned, applications for install or upgrade are queued because they can be processed asynchronously by a Node Manger of the AWS client side processing. Installation processing is a term used for synchronizing the state of the packages that comprise an application with the desired state as described with the application objects and package objects that exist in the configuration repository. Installation processing can occur at multiple times.
As a result of installation processing, the Node Manager may determine whether to install or upgrade an application.
The first part of the processing involves removing unwanted packages in an orderly fashion. The second part involves then upgrading the packages and causing the appropriate events to occur so that the scripts can be triggered. The third part of the processing involves installing the “missing” packages, for example as occurs with either an install or upgrade.
Specifically, in step 3201, the routine looks to see if the USER portion of the packages associated with the application instance has been installed. If not, it continues in step 3204. Otherwise, the routine builds a list of unwanted packages (step 3202), and removes the installation database information for the unwanted packages (step 3203).
In step 3204, the routine looks to see if the SYSTEM portion the packages associated with the application instance has been installed. If not, it continues in step 3209. Otherwise, the routine builds a list of unwanted packages (step 3205); removes unwanted (SYSTEM and USER) packages triggering the PRE_SYSTEM_REMOVAL and POST_SYSTEM_REMOVAL events (step 3206); and builds a list of packages to upgrade (step 3207). In step 3208, the routine upgrades (reinstalls) the SYSTEM and WSM features and triggers the PRE_SYSTEM_UPGRADE and POST_SYSTEM_UPGRADE events.
In step 3209, the routine looks to see if the USER portion of the application instance has been installed. If not, it continues in step 3212. Otherwise, routine builds a list of packages to upgrade (step 3210) and upgrades (reinstalls) the USER feature, triggering the PRE_USER_UPGRADE and POST_USER_UPGRADE events. Note that the USER portion is installed after the SYSTEM and WSM portions.
In step 3212, the routines then adds missing (not yet installed) USER and SYSTEM packages as described below with respect to
More specifically, the Node Manager engages in an orderly process to process application objects one at a time. For each object, NM first checks to see if any dependent application objects need to be processed, and if so, recursively processes these (step 3301).
After this, the NM checks to see if any new packages need to be installed. If they do, the packages are installed as follows:
If any of these steps fail (not shown), portions may be backed out. In particular if the POST_USER_INSTALL script fails, then the Node Manager typically removes the User feature. The User feature usually contains the shortcuts etc. of the local application—therefore this implies that if the post install script fails, the application is not apparently available to the user.
The result of all of this processing is recorded in the installed database on the computing device where the processing occurred. For removals, the relevant package records are removed. For upgrades, fresh information is recorded. For installs, new records are added.
In addition to resolving entitlements, and installing applications and customizing them through the use of scripts, the AWS offers tools to remotely monitor data, state, and status of managed computing devices and terminal servers. For example, the AWS Manager Workbench (see
A Node manager can also interact with the native operating system's database of installed applications.
A Kernel Manager component, briefly described above, typically resides on the Terminal Server or on the managed computing device and can trace all the system calls it intercepts in order to achieve behavior modification (it runs in “kernel” mode, which is privileged like a device driver). A Node Manager can retrieve this trace information and send it to the AWS Manger Workbench for further analysis.
A trace is capable of showing:
File/Directory/Network share accesses
Registry Accesses
Process Start/Stop
Module Load/Unload
Device Driver Load/Unload
Network listens
Network connects
Network disconnects
Other accesses could be similarly incorporated.
One or more of the components of an example AWS can be implement on a general purpose computing system.
In the embodiment shown, computer system 3700 comprises a computer memory (“memory”) 3701, (optionally) a display 3702, one or more Central Processing Unit (“CPU”) 3703, and (optionally) Input/Output devices 3704. The Application Workspace System (“AWS”) 3710 is shown residing in memory 3701. The components of the Application Workspace System 3710 preferably execute on CPU 3703 and manage the generation and use of a virtual workspace and configured and personalized applications, as described in previous figures. Other downloaded code 3730 and potentially other data repositories 3720, also reside in the memory 3710, and preferably execute on one or more CPU's 3703. In a typical embodiment, the AWS 3710 includes one or more Node Managers 3711, one or more configuration repositories 3712, one or more user interfaces 3713 and session support 3714.
In an example embodiment, components of the AWS 3710 are implemented using standard programming techniques. However, a range of programming languages known in the art may be employed for implementing such example embodiments, including representative implementations of various programming language paradigms, including but not limited to, object-oriented (e.g., Java, C++, C#, Smalltalk), functional (e.g., ML, Lisp, Scheme, etc.), procedural (e.g., C, Pascal, Ada, Modula), scripting (e.g., Perl, Ruby, Python, etc.), etc. In addition, the various components of the illustrated embodiments may be implemented by way of a single monolithic executable running on a single CPU computer system, or alternately decomposed using a variety of structuring techniques known in the art, including but not limited to, multiprogramming, multithreading, client-server, or peer-to-peer, running on one or more computer systems each having one or more CPUs.
In addition, programming interfaces to the data stored as part of the AWS process can be available by standard means such as through C, C++, C#, and Java APIs; libraries for accessing files, databases, or other data repositories; through scripting languages such as XML; or through Web servers, FTP servers, or other types of servers providing access to stored data. The configuration repository 3713 may be implemented as one or more database systems, file systems, LDAP servers or any other method known in the art for storing such information, or any combination of the above. In addition, some operations may be implemented as stored procedures, or methods attached to configuration “objects,” although other techniques are equally effective.
Also the example AWS 3710 may be implemented in a distributed environment that is comprised of multiple, even heterogeneous, computer systems and networks. For example, in one example embodiment shown below in
In example embodiments, these components may execute concurrently and asynchronously; thus the components may communicate using well-known message passing techniques. Equivalent synchronous embodiments are also supported by an AWS implementation. Also, other steps could be implemented for each routine, and in different orders, and in different routines, yet still achieve the functions of the AWS.
The Entitlements Portlet 3821 communicates with Information Services 3824 to obtain information from the configuration repository 3841 stored and managed by one or more LDAP Servers 3840. An AWS provided replication mechanism that uses XSLT stylesheets to perform the replication is described further below. This technique is used to replicate the LDAP directories 3841 as needed by the system. The Entitlements Portlet 3821 performs the entitlement resolution process as described in detail above.
The client computing device 3801 uses AWS security services 3830 to implement tunneling for secure connections to the applications servers and to perform remote application processing. The tunnel client also known as the CAP, can be seen in the workstations on the diagram. The components of the security services 3830 also provide Kerberos security services to the system.
XSLT Based LDAP Synchronization
The XSLT based LDAP Synchronization technology (“XLDAPS”) allows synchronization of LDAP data between directories of different types using a unique approach of a transformed replication changelog. Normally LDAP replication is used to replicate directory data between two identical directory server implementations. In some environments, including the AWS described above, every data entry added/deleted/modified in the LDAP is automatically replicated to the other directories. The XLDAPS technology uses XSLT transformation rules to synchronize LDAP changes to different LDAP Directories which may have a different LDAP Schema and a different LDAP Directory Information Tree structure.
A transformation expressed in XSLT describes rules for transforming a standard replication log into a synchronization log. A transformation expressed in XSLT is called a stylesheet. Appendix B is an example XSLT document as used in the AWS to perform LDAP replication.
The XLDAPS technology is implemented as an Directory Server plug-in and is used on an LDAP Directory Server (e.g., Sun Java Directory Server, IBM Tivoli Directory Server, Microsoft Active Directory etc.). It allows add/modify/delete operations performed on one LDAP Directory to be synchronized using corresponding operations on a completely different type of LDAP Directory.
For more information on LDAP—Lightweight Directory Access Protocol—IETF standard—see RFC 3377 http://www.ietf.org/rfc/rfc3377.txt. For more information on XSLT see XSLT—Extensible Style Language Transformations—W3 standard—see http://www.w3.org/TR/xslt.
Multiple instances of the Plugin can run each outputting a different format changelog (a changelog contains the datastream to be replicated to the target directory server) each described by the XSL/XML definition files. More about XSL is found at http://www.w3.org/Style/XSL/
Typical uses of this plugin include the generation of an OpenLDAP style slurpd changelog to replicate LDAP changes to other LDAP servers including Microsoft Active Directory. The XSL stylesheet defines the slurpd replication changelog format and defines the mapping from LDAP attribute types to Active Directory attribute types. This means that the plugin does not need any knowledge of the slurpd changelog format or of the specific details of mapping LDAP attribute types to Active Directory attribute types. Rather, the plugin can just invoke an XSLT processor to read and execute the transformations in order to produce the changelog.
This plugin can also be used to output a changelog for use by other replication and synchronization services such as for replicating changes to an RSA/Ace SecurID server or for producing an audit log in a specific format.
The plugin is designed to work with Sun ONE Directory Server, but could fairly easily be ported to work with IBM Directory Server on Intel or zSeries hardware. Currently it is supported on Intel Red Hat 7.3 Linux working with Sun One Directory Server 5.1.
This plugin is extensible. By using XSL as the changelog format transformation mechanism, new replication requirements can be met without the need to make changes to the software. Adding a new replication log, or changing the attribute mappings etc. can all be done by modifying or creating a new XSL stylesheet and XML file definition.
In an example embodiment, the plugin is called libpaerep-plugin.so and is a re-entrant shared object library. It is configured in LDAP to automatically start when the LDAP processes ns-slapd start.
When the plugin is started, it parses the XML and XSL files. If they are not readable, or they contain syntax errors, the processing aborts and the plugin will not initialize. Otherwise the XSL file is read to determine the list of XSL parameters that are defined.
For efficiency, the XML and XSL files are only read and parsed at initialization time. If the contents of the files are changed, it is necessary to restart the LDAP processes ns-slapd. This is acceptable as the contents of the XML and XSL files should rarely change.
Parameters configured in LDAP are passed to the plugin at startup to determine the format of the changelog. They are defined in order as follows:
The remaining parameters represent a list of LDAP attribute types for which values are always passed as parameters to the XSLT processing after a modify operation. Usually after a modify, only the attribute type values that have been modified are passed together with certain fixed ones (see below).
The following LDAP entry shows an example plugin configuration used for Active Directory Replication.
LDAP Add
After an LDAP entry is added, each added LDAP attribute type is compared with the parameter list specified in the XSL stylesheet. If there's a match, the XSLT Processor is run passing all available parameter values from the new LDAP entry.
In addition to LDAP attribute types, the built in parameters changetype, dn, modifiersname and time are also passed if they are defined in the XSL. See below for a description of these built in parameters.
LDAP Modify
The XSLT processing for an LDAP Modify is similar to LDAP Add. The difference is that only the attribute types that have actually changed are passed as parameters. This is done for efficiency. If an attribute type value changes that is not defined as a parameter in the XSL stylesheet, XSLT processing will not be performed.
XSLT processing is performed once for each attribute type modified.
In addition to LDAP attribute types, the built in parameters changetype, dn, modifiersname and time are also passed if they are defined in the XSL. See below for a description of these built in parameters.
Additional attribute types can be passed after an LDAP Modify even if those attribute types have not changed. This is sometimes required for LDAP Modify XSLT processing. These additional parameters are specified as plugin startup parameters 4 and beyond. See above.
LDAP Delete
The XSLT processing for LDAP Delete is very simple as no LDAP attribute type values are available after the entry has been deleted. The only parameters available are the built in parameters changetype, dn, modifiersname and time. These are passed if they are defined in the XSL. See below for a description of these built in parameters.
LDAP MODRDN
This plugin doesn't support the MODRDN LDAP operation. It may be supported in a future release.
XSL Parameters
Any LDAP attribute type such as uid, cn, mail etc. can be used as a parameter in the XSL stylesheet.
This plugin also allows some built in parameters to be used. These are as follows:
Any parameters passed to the XSLT processing are automatically decrypted if they have been encrypted using the encryption routines as supplied with the Propero workspace software. See the XSL stylesheet example below for details of how these parameters are used.
XSLT Processing
The plugin uses the C API to the Sablotron library as the XSLT processor. This is maintained by the Ginger Alliance Open Resource Centre. See http://www.gingerall.org/charlie/ga/xml/p_sab.xml Sablotron provides access to a set of standard Document Object Model (DOM) functions which are used to parse the XSLT stylesheets.
It is linked with the static Sablotron libraries to avoid a library function clash with an LDAP Function. This means that the plugin is not dependent on the Sablotron shared object libraries.
Message Based Load Balancing
One of the techniques provided by the Application Workspace System is a message bus, which it uses for operations such as load balancing among terminal servers and for security related resources. A message bus is a communications abstraction (provided by code) which allows a requester (or any type of querying agent) to send a broadcast message asking for certain information or requesting a particular response, Servers that can potentially provide the information or the requested service, subscribe to the messaging service and the particular messages that are relevant to the information or services provided by the servers. When the requester broadcasts a request, only those servers that can/desire to provide the information or service respond. In that way, the requestor knows what servers are available and/or appropriate to the desired information or service.
Message based load balancing keeps the application logic of resource selection simple, and also inherently provides up-to-date information that does not include information about resources that are not present or not responding, and thus provides a basis upon which to construct highly-available systems.
For example, the AWS uses message-bus based load balancing to determine which servers to use to launch remote application access requests. This allows the AWS to avoid use of central component to respond to load balancing queries, to ensure the receipt of updated information regarding available servers, and to avoid the complexities of detecting server unavailability, timeouts, etc. For example, the AWS session management server components (such as Session & Connection Control 312 in
Thereafter, servers, such as servers 3903, 3904, and 3905, respond to the particular request. The querying agent 3902 then receives all the possible responses to the original message within a defined timeout period (controlled, for example, by the code invoked by the querying agent to broadcast the message). This timeout period is relatively short, for example 10 seconds, and as such serves to filter out possible respondents that are not in a position to respond—either due to a failure, or due to overloading. In one embodiment, this timeout period is defined as the time period for all responses to be received, and not as a timeout per possible respondent.
The querying agent 3902 then applies an algorithm to the elements of data returned from all the respondents to decide which is the most appropriate respondent. In the case of a query to figure out where to run or access an application or data, the algorithm may be a load balancing algorithm.
Note that the message bus mechanism provided by the AWS also can be used for other purposes, such as voting and consistency checking.
Example pseudo-code of how the message bus is used to perform a load balancing algorithm across Terminal Servers for session management is included as Table 2, below. Note that the parameter “serversToQuery” is a list of possible servers that has been retrieved from LDAP. The load balancing algorithm employed in this code sorts the responding servers based upon statistical information and then chooses which one to use based upon the sort. For example, in the code shown below, the statistics include the number of current sessions being serviced, the percentage of memory being used, and the average processor time being utilized. Other load balancing algorithms can be similarly incorporated.
The above code shows that this is a Session Manager implementation in that it is implementing the interfaces of the session controller for Terminal Servers (e.g., PAESessionImp class is an implementation of the PAESession interface). Other algorithms for load balancing can be similarly defined and integrated in an example AWS.
The underlying support for a message bus can be provided by any compliant message bus provider. In the example embodiment used to support the AWS, which is JAVA based, the message bus is a JMS message bus, for example, provided by MQSeries. Additional information on JMS message buses can be found in http://java.sun.com/products/jms/, which is incorporated herein by reference.
In addition to using a message bus for session management, the message based load balancing techniques can be used to determine which server to use for implementing secure VPN services (such as the tunnel services offered by the Security, Authentication and Other Support Services 314 in
Using message bus load balancing, the AWS web services can select an appropriate tunnel server at the moment it is needed to provide the conduit for secure communications between an AWS client and the AWS Web services.
Grid Location Services
Grid Location Services (“GLS”) represent an integration of the multicast DNS (mDNS) technologies with DNS technology to achieve transparent configurability of servers.
The AWS needs to support the ability to services to be added, removed and to fail over without reconfiguring all the entities that use those services. The AWS implements a mechanism referred to as Grid Location Services to provide this capability. In older systems, mDNS or other broadcast/multicast directory name service systems were used through a network to achieve transparent reconfigurability, but the implementations of firewalls in modern networks prevent such service systems from being effective. There are only a few services that all firewalls allow through, and DNS (Domain Name Service) is one of them.
In a DNS system, names such the server names contained in urls (also known as fully qualified domain names) are mapped to network addresses analogous to how a phone book resolves names to phone numbers but with (typically) multiple steps of translation for all portions of a network address. The directory, or portions of it, are typically replicated throughout a networked environment, but when a change to a mapping of domain name to network address occurs, this change needs to be propagated to the DNS servers that implement the directory. Thus, when a server address fails, or when the network is configured there is a delay to await the change taking effect. In addition, many small local networks do not support a DNS server, and so use other ways to map the address to a final destination—a node on the local network. In addition, as DNS does not check the state of any addresses that it provides (as downstream or caching servers may be providing the address), there is also the possibility that the final hop to a physical machine on a local network will not properly result in a working destination.
mDNS protocols offer a way to provide such mappings without the use of a DNS. Thus, smaller or localized networks may implement mDNS to figure out which computing resource is response to a service request. In mDNS, a request for a service is reconciled with the provider of a service using an IP multicast protocol without a DNS server being implemented. When devices fail, or are reconfigured, they simply don't respond, or respond as appropriate to the service request.
As mentioned, many of today's firewall systems do not allow mDNS messages to pass through the firewall. The AWS Grid Location Services address these problems by allowing remote computers and LANs to use a DNS name and the DNS protocol to resolve a name to an IP address, even though the “last hop” of the DNS resolution process is actually achieved using mDNS. Thus, the devices can be reconfigured at will locally (or even fail) transparent to requesting systems that access them remotely.
Technically, Grid Location Services is a combination of DNS and mDNS. mDNS is used to locate services on the same LAN, and a DNS to mDNS bridge allows entities not resident on the same LAN to take advantage of the capabilities of mDNS to locate services on that LAN.
Note that this is not an implementation of what is sometimes termed “Dynamic DNS” in that the DNS servers in this invention are not updated. Rather, they always return a 0 “time to live” for all responses so they are always re-queried, and each time they will use the state of the MDNS system to resolve the DNS query.
In an example embodiment of the AWS Grid Location Services, the GLS subsystem (the DNS to mDNS bridge), for example, bridge server 4021 in
The GLSMDNS process registers as an MDNS node with a configurable domain and for a number of different service types (see below for example usage), and listens for other registrations in that domain. When other registrations are received, it updates the registration file—this in essence ensures that the registration file contains all of the names and associated IP addresses of nodes registered in a particular domain providing a particular service. In addition, when nodes are no longer present, their information will be removed from the registrations file by the GLSMDNS process.
GLSMDNS can also be extended with particular plug-ins that can check whether a server that is providing a service type is actually providing that service. For example, if the GLSMDNS process is monitoring servers that register LDAP directory services, then GLSMDNS will have a plug-in that will periodically perform LDAP queries against all the servers in the registration file, and for those that do not respond correctly, not include that node in the list of addresses that correspond to that service type in the registration file. mDNS uses a convention of underscores in service names—GLSMDNS strips these off when creating DNS names that map to service names.
As an example of the use of AWS Grid Location Services, suppose 4 mDNS nodes exist in the .local. domain as follows:
In this example, there are a number of different service types, including LDAP services, JMS services, terminal services, and http services. GLSMDNS may have a plug-in to monitor each service type, but LDAP is the one that is commonly used in an example AWS. GLSMDNS monitors mDNS messages for the .local. domain and having received registrations for the above 4 nodes will write out a registration file that includes the following contents:
# Hosts
PC01=192.168.100.41
PC02=192.168.100.42
PC03=192.168.100.43
PC04=192.168.100.44
# Services
LDAP=192.168.100.41, 192.168.100.42
JMS=192.168.100.42
HTTP=192.168.100.43
RDP=192.168.100.44
In this example, GLSDNS is configured to be the authoritative DNS server for the domain WORKSPACE.INT. GLSDNS responds to queries for names in the WORKSPACE.INT domain by using the registration file created from the mDNS information.
As a result, a DNS query for pc01.workspace.int would get the response 192.168.100.41, and a DNS query for http.workspace.int would get 192.168.100.43. GLSDNS uses each service address it finds in turn, so that a DNS query for ldap.workspace.int may return 192.168.100.41 or 192.168.100.42.
This last example demonstrates how the monitoring plug-ins operate—if the GLSMDNS monitoring plug-in for LDAP failed to retrieve a correct LDAP response to the test query, it would remove that node from the LDAP list thus ensuring that clients who are requesting the service LDAP will only receive the IP address of a working LDAP service.
By combining mDNS and DNS, a number of features of both are no longer available—e.g., mDNS has a facility for providing the port number of a service as well as the address, and DNS has a strong focus on caching results to lower network traffic. Neither of which are available using GLS.
However, the advantages of using the GLS approach are several. Servers and services can be added to an embodiment of the AWS without the need for the configuration that DNS systems require and the AWS can dynamically address whether those servers and services are present or not. Also, service queries can take into account whether the service is actually functioning—this is an important part of high availability. Moreover, the GLS functionality is available to any client systems that use DNS without the requirement for additional client side software.
Additional information on DNS is provided in the DNS Specification RFC 1035: http://www.ietf.org/rfc/rfc1035.txt. Details on using DNS to locate services is provided in RFC 2782: http://rfc.net/rfc2782.html. Details on using mDNS are available in the mDNS Specification: http://files.multicastdns.org/draft-cheshire-dnsext-multicastdns.txt.
All of the above U.S. patents, U.S. patent application publications, U.S. patent applications, foreign patents, foreign patent applications and non-patent publications referred to in this specification and/or listed in the Application Data Sheet, including but not limited to U.S. Provisional Patent Application No. 60/752,208 entitled “VIRTUALIZED APPLICATION WORKSPACES,” filed on Dec. 19, 2005, is incorporated herein by reference, in its entirety.
From the foregoing it will be appreciated that, although specific embodiments of the description have been described herein for purposes of illustration, various modifications may be made without deviating from the spirit and scope of the invention. For example, the methods and systems for performing entitlement resolution and packaging discussed herein are applicable to other architectures other than a Windows-based architecture and in non “online” environment. For example, the separation of user administration and application administration can be integrated into a group of peer computer systems connected together. The methods and systems discussed herein are applicable to differing protocols, communication media (optical, wireless, cable, etc.) and devices (such as wireless handsets, electronic organizers, personal digital assistants, portable email machines, game machines, pagers, navigation devices such as GPS receivers, etc.).
Number | Name | Date | Kind |
---|---|---|---|
6301589 | Hirashima et al. | Oct 2001 | B1 |
6625623 | Midgley et al. | Sep 2003 | B1 |
6684331 | Srivastava | Jan 2004 | B1 |
6816882 | Conner et al. | Nov 2004 | B1 |
6922724 | Freeman et al. | Jul 2005 | B1 |
7181436 | Conley et al. | Feb 2007 | B1 |
7222179 | Srivastava et al. | May 2007 | B2 |
7310650 | Felsted et al. | Dec 2007 | B1 |
7464214 | Davies et al. | Dec 2008 | B2 |
7486695 | AbdelAziz et al. | Feb 2009 | B1 |
7526562 | Samprathi et al. | Apr 2009 | B1 |
7571180 | Minyailov | Aug 2009 | B2 |
7660983 | Srivastava et al. | Feb 2010 | B1 |
7743388 | Nikols et al. | Jun 2010 | B2 |
7779091 | Wilkinson | Aug 2010 | B2 |
7831034 | Maximo et al. | Nov 2010 | B2 |
8028026 | Jain et al. | Sep 2011 | B2 |
20020083087 | Deuser et al. | Jun 2002 | A1 |
20020143865 | Tung Loo et al. | Oct 2002 | A1 |
20030014557 | Berger et al. | Jan 2003 | A1 |
20030088587 | Merrells et al. | May 2003 | A1 |
20030091028 | Chang et al. | May 2003 | A1 |
20030140068 | Yeung | Jul 2003 | A1 |
20030144894 | Robertson et al. | Jul 2003 | A1 |
20030172127 | Northrup et al. | Sep 2003 | A1 |
20030191868 | Espino | Oct 2003 | A1 |
20040107125 | Guheen et al. | Jun 2004 | A1 |
20040117667 | Lavender et al. | Jun 2004 | A1 |
20040172558 | Callahan et al. | Sep 2004 | A1 |
20050015507 | Chin | Jan 2005 | A1 |
20050044110 | Herzenberg et al. | Feb 2005 | A1 |
20050044197 | Lai | Feb 2005 | A1 |
20050044356 | Srivastava et al. | Feb 2005 | A1 |
20050091537 | Nisbet et al. | Apr 2005 | A1 |
20050114401 | Conkel | May 2005 | A1 |
20050160172 | Eytchison | Jul 2005 | A1 |
20050267978 | Satapati | Dec 2005 | A1 |
20050283462 | Pennec et al. | Dec 2005 | A1 |
20060031185 | Jose et al. | Feb 2006 | A1 |
20060149760 | Keni et al. | Jul 2006 | A1 |
20060150026 | Kolawa et al. | Jul 2006 | A1 |
20060155773 | Drouet et al. | Jul 2006 | A1 |
20070115282 | Turner et al. | May 2007 | A1 |
20070226753 | Nikols et al. | Sep 2007 | A1 |
20070239859 | Wilkinson et al. | Oct 2007 | A1 |
20070282953 | Jain et al. | Dec 2007 | A1 |
20070283154 | Zhang et al. | Dec 2007 | A1 |
20080034071 | Wilkinson et al. | Feb 2008 | A1 |
20080046576 | Wahl | Feb 2008 | A1 |
20080077752 | Kinoshita | Mar 2008 | A1 |
20080091812 | Lev-Ran et al. | Apr 2008 | A1 |
20080104215 | Excoffier et al. | May 2008 | A1 |
20080133777 | Wilkinson | Jun 2008 | A1 |
20080147787 | Wilkinson et al. | Jun 2008 | A1 |
20080256090 | Dietterich et al. | Oct 2008 | A1 |
20080263640 | Brown | Oct 2008 | A1 |
20080301258 | Arndt et al. | Dec 2008 | A1 |
20090150398 | Raut | Jun 2009 | A1 |
20090216815 | Braginsky et al. | Aug 2009 | A1 |
20100082979 | Edwards | Apr 2010 | A1 |
20100269160 | Kalonji et al. | Oct 2010 | A1 |
Number | Date | Country |
---|---|---|
2007075846 | Jul 2007 | WO |
Number | Date | Country | |
---|---|---|---|
20080147745 A1 | Jun 2008 | US |
Number | Date | Country | |
---|---|---|---|
60752208 | Dec 2005 | US |
Number | Date | Country | |
---|---|---|---|
Parent | 11642990 | Dec 2006 | US |
Child | 11929302 | US |