Note: Although retrieval warehouse items is distributed,
the same is not true when adding or updating warehouse items.
When adding or updating items to a warehouse, the client must directly
contact the warehouse where the item is to be stored.
A catalog component enables users to find what content exists
throughout the enterprise. Although the content is physically stored
in a warehouse, the catalog stores
catalog items. A catalog item describes one or more
warehouse items. Any
client can issue a query on a catalog to find catalog items.
The catalog item can be used to access the corresponding
warehouse content.
The following are some of the queries available to a client:
The overall catalog architecture can be functionally divided into
four subcomponents: catalog server, catalog storage subsystem,
catalog agent, and catalog peers.
Catalog information is logically represented as catalog items,
which consist of:
Catalog
Catalog Terminology
Catalog Item
Catalog Server
A catalog server accepts and retrieves catalog information for client
applications and other catalog servers (peers).
A catalog server GUI allows a catalog
administrator to configure and run a catalog server.
Item storage and retrieval is handled by the
catalog storage subsystem component.
Each catalog item contained in a catalog is locale specific, but not all catalog items need be for the same locale. Thus a single catalog can accept and retrieve catalog information from clients in any number of locales.
The catalog server GUI also allows the catalog administrator to choose and configure a catalog storage subsystem. See the Storage System Settings section of the catalog configuration for details on how to choose your storage subsystem.
The content of a catalog storage subsystem, irrespective of its implementation, consists of catalog items.
The search distributed to other catalogs is timed; the search will not last forever. Results from this distributed query are returned piecemeal to a "holder" containing all the search results. Any duplicate catalog items will be removed by the holder before the results are returned to the client.
Note: Although searching for catalog information is distributed,
the same is not true when adding or updating catalog information.
When adding or updating information to a catalog, the client must directly
contact the catalog where the information is to be stored.
The Network Deployment Client component also provides additional function, such as
management of catalog and warehouse content ordered or added by the end user. This
avoids multiple copies of the same content in one client environment.
Also, the Client component includes a network class loader, which
Java programs can use to dynamically download and load classes stored as warehouse
content.
To do so, type:
java com.ibm.beans.util.Configure
Using this program, specify the name of the Bean Extender working directory, where various
files used by Network Deployment will be saved. Also, specify the host names of the warehouse
server and catalog server which the client component is to connect to by default. The specific
connection information should be obtained from the server administrator. If the server administrator
specifies default connection values, a user only has to know the host name of the server.
Finally,
the client component uses a daemon process which must be running when any client operations are
performed. In the configuration program, specify the connection information in the manner as
the warehouse and catalog servers. On single user machines, the defaults provides should always
be sufficient. On systems with multiple concurrent users, you can choose a connection name which
will not conflict with other users. The client daemon and client GUI will both use the connection
name you specify.
The warehouse and catalog components each require component-specific setup. For this information,
refer to the following usage sections for the respective components.
The rmiregistry utility does not return to the command line, and must be left running.
The rmiregistry utility does not return to the command line, and must be left running.
To configure and start a catalog type:
The optional "-s" flag causes the catalog server to autostart.
The catalog server GUI appears as shown in the following figure:
When the catalog is brought up for the first time, configuration should be performed.
Configuration consists of three panels that include Status settings, Storage
system settings and Catalog peer settings.
Once the catalog configuration is completed, the catalog can be started or
stopped by using the Start server and Stop server buttons
on the Status panel.
Once the catalog is stopped, the administrator can terminate the catalog server
by going to the Exit panel
and pressing the Terminate server button.
The CatalogInit utility can be used to
easily create part of the catalog hierarchy.
If this is the first time the relational database is used as the
catalog storage subsystem, then check the
Create new catalog tables option. Due to a limitation
of JDBC, you will need to supply the actual SQL name used for
LONGVARBINARY data. The Database entry field can be
used instead, and the customizer will attempt to fill in
the SQL name for that database's LONGVARBINARY data.
The Drop existing catalog tables choice provides a
convenient way to remove all the catalog tables from your database.
It can be used in conjuction with the Create new catalog tables
choice to ensure a totally empty catalog.
Once the catalog server is stopped, the administrator exits the server
by going to the Exit panel and pressing the Terminate server button.
The state information for the catalog server is saved in the file
specified.
To invoke the CatalogInit utility, type:
In the command above, the CatalogInit.data file might contain
something similar to the following:
rmiregistry
The rmiregistry does not return to the command line and must be left running.
To start the client daemon type the following command:
java com.ibm.beans.netdeploy.client.ClientImpl
This program does not return to the command line and must be left running.
Now the client graphical interface can be run. To start this program type:
java com.ibm.beans.netdeploy.client.ClientMain
The panel that appears is shown in the following figure:
The client panels use a "wizard" style approach, so that you are guided through a succession
of panels for the selected operation. You must fill in all necessary data before you can
advance to the next panel.
Adding content to a warehouse or catalog is referred to as stocking, and updating content
is referred to as restocking.
By default, all stock and restock
operations will be issued to the configured default warehouse or
catalog specified in the Network Deployment Setup And Initialization.
section. These defaults can be overridden by Set Warehouse Server Connection and
Set Catalog Server Connection dialogs. These dialogs can be accessed via the Options
menu item. For order operations, the specific warehouse or catalog being connected to is not
important, so long as the warehouse or catalog with desired content can be accessed directly or
indirectly from the server to which you are connected. If it is, desired content is located
and obtained for you irrespective of its particular location.
To exit the client graphical interface, select Exit under the File menu item at any time.
In this panel, enter the name of the content being stocked. Content would typically be a Java
class file or jar file. After the content file name is specified, press the Next button, and
you will be advance to the following panel:
From this panel, you can specify fully qualified class names that can be referenced externally,
similar to exports in libary files. Specifying these names allows the network class loader included
with the Network Deployment client component to locate your content when the exported class is
referenced.
When all of the necessary class names have been entered, press the Next button to advance to
the next panel, shown below.
The above panel allows you to specify dependencies you have on any other warehouse content.
It is important to specify all dependencies so that all dependent content is also downloaded
to a client along with the base content being ordered.
It is assumed that you already ordered the warehouse items upon which you are dependent,
or have published them to the warehouse yourself. In either case, they will appear in the
client registry of warehouse items. To view the registry press the Browse Client Registry
button, and select any dependent items from the ensuing registry list dialog window shown below.
When you have made the necessary selections and exited the dialog panel, press Next to advance
to the next panel, shown below.
At this point, confirm that you want to actually perform the operation by clicking the
button entitled Commit Current Content & Perfrom Operation. Processing will begin as soon as you
press Next. Upon completion of the operation, a panel similar to the one shown below will
appear.
From this panel, press the Next button to return to the main panel from which you began.
To create a catalog listing, select Stock Catalog Content on the main panel and press the
Next
button. This will begin a series of panels for the catalog stock operation. The first panel is shown
below.
From this panel you specify what paths catalog browsers can traverse in order to reach
your content. You can specify new or existing paths. At any point in an existing path, you
can branch off and create new paths. The path elements should represent logical organizations
under which your content can be classified.
To add new path entries, press the Browse Navigator button to bring up the Catalog Navigator
dialog, shown below.
You can browse the catalog to find paths that already define catagories to which your catalog
entry belongs. Using existing paths will make catalog browsing more productive than having
many unique paths that lead only to one piece of content. Enter in the desired path info in the
fields below the tree view. When users browsing the catalog run the mouse button over your entry,
the title is displayed as fly-over info. If the user clicks on an entry, the description information
is displayed. The description can be tailored to the context of the path and should be very brief.
You will have the opportunity to enter more detailed information in a subsequent panel.
When the OK button is pressed from the Catalog Navigator dialog above, you are returned to the
list of specified paths. The panel below shows how the above entry would be listed.
When all of the desired paths have been entered, press to Next button to advance to the next
panel, shown below.
In the panel, enter the catalog item title, description. If you have a web page that you want
readers to visit, enter it in the URL field. In the Contact URL field, enter , e-mail URL,
home page URL, etc., so users can contact you. A catalog entry can be added for each locale desired.
Catalog browsers only see entries for their locale. After this form is completed, press the
Next button to advance to the next panel.
Catalog items typically refer to one or more warehouse items. In the panels shown so far, the
sample entries describe a Jar file. It is assumed that you have stocked (or ordered) the warehouse
item already, which means there is an entry for it in the Client Registry. To select the warehouse
item or items that are to be downloaded when a user orders through this catalog item, press
the Browse Client Registry button. The registry list on your machine should appear similar to
the example shown below.
From this list, select all content associated directly with this catalog item and press the Done
button. The original panel will show the items selected, as in the panel shown below.
Press the Next button to advance to the next panel, shown below.
From this panel, specify attributes that apply to the warehouse content being catalogued.
For example, you can allow catalog browsers to filter out all entries which do not describe
Java components. There is no pre-defined set of attribute-value pairs. However, with
use, a defacto set of commonly used attributes should emerge.
When you have finished entering attributes, press the Next button to advance to the Keywords
panel, shown below.
In this panel, specify all keywords which are relevant to the content being catalogued and
press the Next button to advance to the next panel, shown below.
Catalog items can reference other catalog items. This feature is designed to facilitate
directly navigating from one catalog entry to another related one. An example of how this
might be useful is if the item you are cataloging depends on some other item which is also
catalogued. Creating such a reference would make it easy for catalog browsers to visit
related catalog entries. This should aid them in making order choices. Reference information
is stored to the catalog, but this feature is not
yet being exploited in the client graphical interface.
When finished with this panel, press the Next button to advance to the next panel to
commit the stock operation. From here, proceed as described in the paragraph on the
Perform Current Operation panel shown in the warehouse item stock
sequence described above, until you are returned to the main panel.
The Network Deployment client assumes you have already published or ordered the warehouse
item that you want to restock. This means that an entry for it exists in the Client Registry.
To select the warehouse item to be restocked, press the Browse Client Registry button to
view warehouse entries registered in the Network Deployment client registry. The
client registry should appear in the Client Registry View with items
shown in catalog item stock sequence. From this Client Registry dialog, select the item
to restock and press the Done button to return to the original panel. Then press the
Next button to proceed. From this point on, the sequence is exactly like the warehouse
stock sequence described above, except that all of the values are already filled in. Any
of them can be changed as desired.
Select the desired entry and press the Done button.
Alternatively, you can use the Catalog Navigator to choose the catalog item to restock.
Press the Next button to proceed to the next panel. From this point on, the sequence is
exactly like the catalog stock sequence described above, except that all of the values are
already filled in. Any of them can be changed as desired.
If a warehouse item included as part of an order, either as part of the base order or as a
dependency, is already on the client machine, it will be removed from the total list of warehouse
items to be ordered and will not be downloaded again.
Once the order is complete, and invoice file is stored in the order directory, or appended to the
existing invoice file. This invoice information provides all of the details on what content was
actually downloaded to the client machine, as well as a listing of content that already resides on
the client machine and was thus not downloaded.
The sequence of order panels is described below. After selecting Place Order on the main
panel and press the Next button, you will advance to the following panel.
In this panel, press the Browse Navigator button to view the Catalog Navigator Dialog.
The title of the path being viewed is shown when cursor is over a catalog item icon. For more detail,
click on the item's icon. The information will be shown in the lower portion of the dialog window.
After locating a catalog item of interest, click on it to select it, and press OK. The Catalog
Navigator Dialog is shown below.
The selected catalog item will appear in the originating panel, as shown below.
To continue, press the Next button. The next panel is shown below.
In this panel, specify the subdirectory where you want to place the order. If the content being
ordered consists of Java class files, this directory should be specified in the CLASSPATH.
If the content is packaged in a Jar or Zip file, it is extracted into the specified order directory.
The original Jar or Zip is not deleted, and remains in the order directory.
After specifying the order directory, press the Next button. The next panel appears as shown
below.
In this panel, specify whether you want to subscribe to the content or not. If you subscribe to
the content, it will be updated daily. Content is actually downloaded only if it has changed. For
subscriptions to be processed,
the client daemon must be running. If you do not subscribe, you can always manually initiate updates,
as described later.
After choosing a subscription option, press the Next button. The next panel is the
Perform Current Operation panel that was described in a previous section.
After pressing Next in this panel, the order is processed, and you advance to a panel that shows
the order status. This panel is shown below.
At this point, order processing is complete. Press Next to return to the main panel.
With an ordinary ASCII text editor, you can view the invoice file, located in the specified order
directory, for order details. The name of the file is INVOICE. An example invoice is shown
below.
To udpate a previous order, select Refresh Previous Order from the main panel and press
Next. The first panel is shown below.
From this panel, you press the Browse Client Registry button to view the list of orders that
have been placed previously. This dialog window is shown below.
Select the order to be refreshed and press Done. The original panel is updated with the
order selected, as shown below.
At this point, press Next to advance to
Perform Current Operation panel that was described in a previous section.
Pressing the Next button starts the actual update process. When processing completes, you
advance to the next panel, shown below.
At this point, the order refresh is complete. Press Next to return to the main panel.
The invoice file is appended with the order update details.
An example order refresh invoice listing is shown
below.
Client
While the Network Deployment Catalog and Warehouse components both have client-side
classes, both operate more or less independently. The Network Deployment Client
ties these two together from the end user's perspective in the form of a graphical
front end which can be used to deploy warehouse and catalog content, browse the catalog for
warehouse content, and to order browsed warehouse content. Ordered content can be
subscribed, so that the content is refreshed on a periodic basis. The refresh is
initiated by the Client component. Whenever content is downloaded, all warehouse
content identified as a dependency is also downloaded, if necessary.
Network Deployment Setup And Configuration
As of now Network Deployment software is installed as a part of Bean Extender.
A machine that has the Network Deployment package installed on it can be
be used as a warehouse, a catalog or a warehouse/catalog client, or any combination.
Before starting any Network Deployment component, run the Bean Extender Configuration
program: com.ibm.beans.util.Configure.
Configuration And Usage
Servers and daemons can be started through the
Bean Extender configuration utility or by entering
java com.ibm.beans.netdeploy.Start from a command line once the configuration
has been completed through the configuration utility.
Servers can also be started individually from the command line. This section describes the
latter method.
Warehouse Server Configuration And Usage
It is necessary to start the rmiregistry utility before a warehouse is started.
To do this, type the following command at a command-line prompt:
rmiregistry
To configure and start a warehouse server, type:
java com.ibm.beans.netdeploy.warehouse.WarehouseImpl -s
The optional "-s" flag causes the warehouse server to autostart.
A warehouse customizer will appear, as shown in the following figure:
When the warehouse is brought up for the first time, the server connection
information must be filled in.
Server Connection Information
The server connection settings are needed so that a warehouse can be
bound in the Java RMI registry running on the warehouse server machine.
The host name will automatically be filled in by the warehouse server.
Enter the domain name, server name and port number in the appropriate
fields. Typically, the defaults provided are sufficient. Only if you
run the rmiregistry utility with other than the default port number of
1099 or run multiple warehouse servers on one machine should you need
to deviate from the defaults. The host name, however, must always identify
the local host.
The settings may be also restored from a previous session by pressing
the Read From File... button.
Once the settings have been filled in, press the Apply Settings
button to apply these settings to the warehouse server. To save the
warehouse server settings for use in later sessions, press the
Save To File... button.
Storage System Settings
The storage system, which is a part of the warehouse, performs content
storage and content distribution. Storage System settings allow
warehouse administrator to choose a particular
storage and configure it. By default, the warehouse uses a base storage
system which is built on top of the native flat file system. In this
storage system the warehouse content is stored in separate files and no
authentication or authorization check is necessary to access the content.
When the Next button is pressed, a UI panel for customizing the
selected storage system is displayed in the controller. For example,
the customization for the base storage system looks as follows:
Warehouse Peers
Warehouse Peer settings allow administrators to add or remove peer
warehouses that this warehouse needs to know about. If a client machine
requests certain warehouse content that is not present in the warehouse
storage system, then the warehouse forwards the content request to the
peer warehouses that it knows about. The following figure shows the
configuration of peer warehouses.
Once the warehouse setup is completed, the warehouse can be started or
stopped by using the Start Server and Stop Server buttons
on the main customizer panel. You may also exit the server by using the
Exit Server button.
Catalog Server Configuration And Usage
It is necessary to start the rmiregistry before a catalog is started.
If the rmiregistry has not already been started,
type the following command:
rmiregistry
java com.ibm.beans.netdeploy.catalog.CatalogImpl -s
Status Settings
RMI settings are needed so that a catalog can be bound in the Java RMI
registry running on the catalog server machine. Enter host name,
domain name, server name and port number in the appropriate fields.
Typically, the defaults provided are sufficient. Only if you
run the rmiregistry utility with other than the default port number of
1099 or run multiple catalog servers on one machine should you need
to deviate from the defaults. The host name, however, must always identify
the local host.
Storage System Settings
The storage system settings allow a catalog administrator to
choose a particular storage and configure it.
By default, the catalog uses the PersistentCatalogTableFactory
system which uses the Java persistence model and the native file
system. No additional configuration is needed to use this default
storage system.
If a relational database and its JDBC driver have been installed,
the administrator can choose to use the JDBCCatalogTableFactory
storage system. The configuration GUI for this storage system is
shown below.
Catalog Peer Settings
The Catalog peer settings allow administrators to add or remove peer
catalogs that this catalog needs to know about. If a client machine
searches for catalog information, this catalog will automatically forward
the search request to these peers. Enter the RMI information for each
catalog peer.
Catalog Termination
The catalog server must be stopped before the administrator can exit
from the server. Stopping the server allows any current operations
to complete, but prevents any new client requests from being processed.
To stop the server, press the Stop server
button on the Status panel.
Using the 'CatalogInit' Utility
The CatalogInit utility can be used to
easily create any part of the catalog hierarchy. This utility
uses a file as input. The file specifies the path
information and, optionally, a brief title associated with that path.
java com.ibm.beans.netdeploy.catalog.CatalogInit CatalogInit.data
Hardware (Computer related hardware)
Systems (Complete desktop systems)
Dell (based in Austin, Texas)
Compaq (based in Houston, Texas)
IBM (based in Armonk, New York)
Gateway (based in Souix Falls, South Dakota)
Graphics Cards
Diamond (Monster 3D)
Number 9 (Revolution, Imagine)
Canopus (3D accelarator based on VooDoo 3dfx chip)
ATI (All-In-One, Wonder)
Disk Drives (Fixed and removable)
IOMEGA (Zip, Jaz)
Seagate
Maxtor
Western Digital (Caviar)
Quantum (Bigfoot)
Software (Computer related software)
Applications
Productivity
Word (Microsoft)
Excel (Microsoft)
Access (Microsoft)
WordPro (Lotus)
Paradox (Corel)
Suites
Office (Microsoft)
SmartSuite (Lotus)
WordPerfect Suite (Corel)
Data Bases
Paradox (Corel)
Access (Microsoft)
FoxPro (Microsoft)
DBase (Corel)
Sybase (Oracle)
DB2 (IBM)
Games
Action
DOOM (id Software)
Uprising (3DO/Cyclone)
Business
Monopoloy (Parker Brothers)
Zapitalism!
Simulation
Comanche 3 (TerraNova)
F-22
Strategy
X-Com Apocalypse
Total Annihilation
Compilers
C
Cobol
Fortran
GLUE
Java
Oberon
NetRexx
Pascal
PL/I
Interpreters
Forth
GLUE
Java
NetRexx
Perl
Rexx
Client Usage
Starting the Client Graphical Interface
The client component inclues a "daemon" which must be started and left running while the
client graphical interface is being used. It is necessary to start the rmiregistry before
the client daemon is started. To do this, type the following command:
Stocking Warehouse Content
Adding content to a warehouse is referred to as stocking a warehouse.
To stock a warehouse, select Stock Warehouse Content on the main panel and press the Next
button. This will begin a series of panels for the stock operation. The first panel is shown
below.
Stocking Catalog Content
The catalog provide a means for other potential users to learn about Java components, applications,
or other content that you have deployed to a warehouse. Before reading this section, you should
first read the section, Catalog in the
Overview section to gain an understanding of catalog terms and concepts.
Retocking Warehouse Content
From the Client main panel, select Restock Warehouse Content and press the Next button to
proceed to the first restock panel, shown below.
Restocking Catalog Content
Restocking a catalog item can be done more quickly through the client registry if you have
already published or ordered the catalog item, since a client registry entry for
it exists. To select the catalog item to be restocked, press the Browse Client Registry button
to view catalog entries registered in the Network Deployment client registry. A view of the
catalog entries in the Client Registy is shown below.
Ordering Warehouse Items
With the Network Deployment graphical interface a user can browse the catalog and order any
warehouse content listed. When an order is placed the catalogued warehouse item or items are
downloaded to the client machine to the specified subdirectory, as well as all warehouse items
identified as dependencies. This includes dependencies of dependencies, and so on. All of the
content downloaded to the client machine is recorded in the Client Registry.
****************************************************
Invoice for new order
Order Title: Bert's Buttons
Order subdirectory: /u/wme/Orders
Time: 21-Nov-97 4:56:26 PM
****************************************************
Items ordered
-------------
Name: buttons.jar
New items downloaded
--------------------
Name: utils.jar
Item ID: 7f2b9d96679ac243bd2de6e0664f5e1b
Item type: dependency
Name: buttons.jar
Item ID: 2938a9dcfc4bce7dceec0378fdf937d8
Item type: base item
Items already installed (not downloaded)
----------------------------------------
None
Refreshing Previously Ordered Warehouse Items
Updates to any previous order can be initiated manually. It does not matter whether a subscription
was placed on the order. Content is actually downloaded only if it has changed since the last
download. All dependencies of the base order are included in the order referesh. When the referesh
is complete, the INVOICE file is appended with details, such as what new files were downloaded,
what files were updated, and what files are already up-to-date.
****************************************************
Invoice for refreshed order
Order Title: Bert's Buttons
Order subdirectory: /u/wme/Orders
Time: 21-Nov-97 5:04:57 PM
****************************************************
Items ordered
-------------
Name: buttons.jar
New items downloaded
--------------------
None
Items updated
-------------
Name: utils.jar
Item ID: 7f2b9d96679ac243bd2de6e0664f5e1b
Item type: dependency
Pathname: /u/wme/Orders/utils.jar
Items up-to-date (not downloaded)
---------------------------------
Name: buttons.jar
Item ID: 2938a9dcfc4bce7dceec0378fdf937d8
Item type: base item
Pathname: /u/wme/Orders/buttons.jar
[ Top of Page | Previous Page | Next Page | Table of Contents | Documentation Homepage ]