0% found this document useful (0 votes)
7 views20 pages

Box Classifier Integration Guide

The Box Classifier Guide provides detailed instructions for installing and configuring the Box Classifier application, which integrates with the Box content management platform to manage document classifications. Key features include monitoring events, synchronizing classifier labels with Box metadata, and enforcing security policies for classified documents. The document outlines steps for initial setup, policy control, and operational functionality within the Box environment.
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
7 views20 pages

Box Classifier Integration Guide

The Box Classifier Guide provides detailed instructions for installing and configuring the Box Classifier application, which integrates with the Box content management platform to manage document classifications. Key features include monitoring events, synchronizing classifier labels with Box metadata, and enforcing security policies for classified documents. The document outlines steps for initial setup, policy control, and operational functionality within the Box environment.
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd

Box Classifier Guide

UM643301
November 18

© Boldon James Ltd. All rights reserved.


Customer Documentation
This document is for informational purposes only, and Boldon James cannot guarantee the precision of any information supplied.
BOLDON JAMES MAKES NO WARRANTIES, EXPRESS OR IMPLIED, IN THIS DOCUMENT.
Box Classifier Guide UM643301

Contents
1 Introduction ........................................................................................................................................... 3

2 Installation and Deployment options.................................................................................................. 3


2.1 Initial Configuration of the Box Enterprise ................................................................................. 3
2.1.1 Create a Developer Account 3
2.1.2 Set up Two-Factor Authentication 3
2.1.3 Create the Box Classifier App 3
[Link] Basic Configuration ...................................................................................................... 4
[Link] Public Key Management .............................................................................................. 4
2.2 Authorise Box Classifier for Your Enterprise ............................................................................. 5
2.3 Create the Box Metadata Classification Policy .......................................................................... 5
2.4 Installation of Box Classifier ...................................................................................................... 6
2.5 Configuration of Box Classifier .................................................................................................. 7
2.6 Policy control ............................................................................................................................. 9
2.7 The Box Classifier Service ........................................................................................................ 9
3 Functionality ....................................................................................................................................... 10
3.1 Events and Actions .................................................................................................................. 10
3.2 Uploading a document to [Link] ......................................................................................... 11
3.3 Uploading a new version of a document to [Link] .............................................................. 11
3.4 Synchronisation of documents ................................................................................................ 11
3.5 Adjusting the Classification of a document in [Link] ........................................................... 11
3.6 Unlocking a document in [Link] .......................................................................................... 11
3.7 Copying a document in [Link] ............................................................................................. 12
3.8 Restoring a document from Trash in [Link] ........................................................................ 12
3.9 Preventing documents with specific Classifier labels being retained in [Link] ................... 12
3.10 Quarantining documents with specific Classifier labels in [Link]........................................ 12
3.10.1 Changing an Adjudicator 13
3.11 Using Box Notes in [Link] ................................................................................................... 13
4 Supporting Users with Languages other than English .................................................................. 14

5 Load Balancing ................................................................................................................................... 15

6 Audit Events ........................................................................................................................................ 16

7 Interworking with Classifier applications ........................................................................................ 18

8 Deployment and Infrastructure Considerations .............................................................................. 19


8.1 Firewalls and Proxies .............................................................................................................. 19
8.2 Load and Performance ............................................................................................................ 19

[Link] 2
Box Classifier Guide UM643301

1 INTRODUCTION
This document summarises the features supported by Box Classifier.
Box Classifier is an integration application that controls documents labelled using Classifier within the Box
([Link]) environment. Box is a content management platform that helps people securely store, access,
share, and collaborate on files across devices. The Box platform does not recognise the Classifier labels that
may be attached to documents. Instead, it uses its own metadata system for classification and access control.
Box Classifier automatically synchronises Classifier labels with the Box metadata system for classification.
The main product features are:
 Monitor events within the customer’s Box Enterprise.
 For document uploads synchronise Classifier labels with the Box metadata system for classification.
 Optionally delete or quarantine documents with specific Classifier labels.
 Optionally scan existing documents within the customer’s Box Enterprise before monitoring new
events.

2 INSTALLATION AND DEPLOYMENT OPTIONS


2.1 Initial Configuration of the Box Enterprise
Box Classifier is a custom application that uses Box Platform. To enable Box Classifier to work with your
Box Enterprise you will need to create a configuration for the custom application recognised by Box and then
authorise those connection details within your Box Enterprise. Follow the instructions below to create the
initial configuration and authorise Box Classifier for your Box Enterprise
The following instructions summarise the guidelines provided by Box for creating this initial
configuration. Administrators should review the official Box Guidelines at
[Link] before continuing.

Please note that these instructions and any screenshots are correct at time of writing but may change over
time. For the definitive guide on configuring an App in [Link] refer to the Box documentation as it
appears on their website.

2.1.1 Create a Developer Account

To create the initial configuration for Box Classifier using Box Platform, you will first need to access the Box
Developer Console.
You can sign-up to create a developer account at: [Link]
If you already have an account, use: [Link]

2.1.2 Set up Two-Factor Authentication

Set up two-factor authentication as follows:


 Log in to Box, within your Account Settings, select Security.
 Check Login Verification and follow the instructions to enable 2-step verification.
2.1.3 Create the Box Classifier App

Use the Box Developer Console to create your new app.

[Link] 3
Box Classifier Guide UM643301

 Log in to Box and select Dev Console.


 Select Create New App.
 Select the Enterprise Integration app type.
 Select the OAuth 2.0 with JWT (Server Authentication) option.
 Choose a unique name for the app, such as Classifier Integration for [My Company Name].
 Press Create App.

[Link] Basic Configuration

Ensure that your new App has the following options set:

 Authentication Method
o OAuth 2.0 with JWT (Server Authentication)
 Application Access
o Enterprise
 Application Scopes
o Read and write all files and folders stored in Box
o Manage users
o Manage enterprise properties
 Advanced Features
o Perform Actions as Users
o Generate User Access Tokens
[Link] Public Key Management

Box Classifier makes requests to the Box Cloud Storage System that are signed using a private key and are
verified by the public key registered with the Box System.
Generate the RSA key pair using the [Link] toolkit.
The OpenSSL project does not distribute any code in binary form, and does not officially recommend any
specific binary distributions. An informal list of third party products can be found on
[Link]
During production, Boldon James used the [Link] package to generate the OpenSSL RSA
key pair with the commands below.
Generating a Private Key
openssl genrsa -aes256 -out private_key.pem 2048

Generating a Public Key


openssl rsa -pubout -in private_key.pem -out public_key.pem

[Link] 4
Box Classifier Guide UM643301

Submitting the Public Key


Once you have generated a public key, you will need to save it in your Box Classifier application
configuration under Add and Manage Public Keys.
Further advice on generating the RSA key pair can be found at:

[Link]
keypair

Note: Be sure to protect your private key, as this is the only means to verify your application's identity.

2.2 Authorise Box Classifier for Your Enterprise

Grant authorization for your Box Classifier application in the Box Enterprise Admin console.
 Log in to Box and select Admin Console.
 Go to the Enterprise Settings and click on Apps.
 Within the Custom Applications section, choose Authorize New App.
 Enter the API Key for your Box Classifier Application. The API Key is the Client ID (OAuth 2.0
Credentials) of your app, see Dev Console | Classifier Integration for [My Company Name] |
Configuration | OAuth 2.0 Credentials | Client ID. Entering the API Key will display the
requested Applications Scopes and User Access Level that determines which users the application
has control over.
Once authorized, the Box Classifier Application will be enabled within your Enterprise.

2.3 Create the Box Metadata Classification Policy


Security classification is available to all Box Governance customers. If you don’t see the classification options
contact [Link] to ask them to enable the Box Governance Service for your Enterprise. Security
classification is a feature that enables customers to classify files based on their confidentiality and enforce
security policies associated with that confidentiality level.
To create a new classification using the Classic Admin Console:
 Log in to Box and select Admin Console.
 Go to Policies and select Content Controls.
 Click on Create.
For the New Admin Console the option is located at Governance/Classification
Now complete the following steps:
1. Begin by naming your classification label. This name must be unique, and have a
maximum of 40 characters.
2. You also have the option to apply shared link restrictions on classified files. Select
one of the following three levels of shared link restrictions available:
 Public - No restriction on sharing the classified file
 Company and Collaborators only - No external sharing of the classified file
[Link] 5
Box Classifier Guide UM643301

 Collaborators only – company-wide or external sharing of the classified file


is prohibited.
3. Once you are finished modifying your classification policy settings, click Create.
4. You cannot rename a classification once it has been created, and existing
classifications cannot be deleted. Confirm that you understand these restrictions,
then click Create to proceed.

For example:

Figure 1: Example Box Classifications

2.4 Installation of Box Classifier


If there is an existing installation of the software, uninstall that version of the software before proceeding.
The product is installed into the server environment by running [Link] from the Classifier Bundle
under the Box Classifier subfolder. The installation program presents a sequence of standard setup screens:

 A Welcome screen, where the Next button should be selected.


 An End-User License Agreement screen. Carefully read and accept the agreement then
select the Next button.
 A Select Installation Folder screen. Confirm the installation folder by selecting the Next
button.
 A Ready To Install screen, where the Install button should be selected.

The following steps are required to establish a working Box Classifier installation:

[Link] 6
Box Classifier Guide UM643301

 Apply your Box Classifier licence to your Classifier Configuration using Classifier Administration
Global Settings / Licence mechanisms.
 Configure Box Classifier Policy control (Rules).
 Ensure that the updated configuration is Published (using Classifier Administration / Publish) to a
location accessible to your intended Box Classifier server(s).
 Ensure that the relevant Classifier configuration registry keys are set on the server that is to run Box
Classifier. For example, if the Classifier Configuration is published via File Store, then set up the
following registry entries with the relevant site values:
o HKEY_LOCAL_MACHINE\SOFTWARE\Boldon James\ConfigManager\ServiceMode
 ServerRootType REG_DWORD 0
 ServerFileSystemRoot REG_SZ \\ICS-SVR-1\Classifier Configuration
 LabelConfiguration REG_SZ Classifier Test
 Policy REG_SZ All users
o (You can find comprehensive details of the Classifier registry search algorithm in the
Classifier Administration Guide Configuration Deployment section).
 Complete the Configuration of Box Classifier.

2.5 Configuration of Box Classifier


After creating the initial configuration within Box and authorising that configuration for use with your Box
Enterprise, configuration of the Box Classifier product can continue.

Note: For Box Classifier to connect to your Box Enterprise the UTC time on the machine must be
synchronised with the Box servers. Check and if necessary synchronize the Box Classifier machine’s time
with a trusted, external resource. A difference of a few seconds in the times is allowed.
Box Classifier connects to your Box Enterprise through the two endpoints of [Link] and
[Link]. You may need to adjust Firewall configuration and Proxy Server configuration to allow this
connection. See Section 8.1 for more detail.
Use the Box Classifier Configuration Console that is installed into your server’s Apps repository to finalise
configuration of Box Classifier.
The Configuration console, as shown in figure 2, writes configuration data to the application folder on the
server where you deploy the Box Classifier.

[Link] 7
Box Classifier Guide UM643301

Figure 2: Box Classifier Configuration Console

Using the initial configuration created in section 2.1.3:


 Enter the Client ID, Client Secret and Public Key ID.
 Enter the Private Key and Private Key Password associated with the Public Key created in section
[Link]
 Enter the Enterprise ID found in the Account Info page of your Box Enterprise Settings.
The Adjudicator Name field identifies the designated administrator or adjudicator that can act when there is
a mismatch between classifier labels and box classification values. Click on the Select… button to show the
Enterprise Users that Box Classifier can access.

[Link] 8
Box Classifier Guide UM643301

The Filter for the Users Handled is only used for load balancing. Load balancing is only necessary if your Box
Enterprise has a large number of users or usage is high. If you need to employ load balancing see section 5
The Scan All Files Before Monitoring New Events indicates whether existing files in the Box enterprise
should be analysed for label inconsistencies. The request to scan all files is valid for one hour once the
Request button is clicked and the configuration changes are committed using the OK button. The scan will
take place when the Box Classifier Service is re-started. See The Box Classifier Service below for more details.

2.6 Policy control


Box Classifier operates using the Classifier Policy configured for the local machine. Refer to the Classifier
Administration Guide  Rules Supported for details on creating rules appropriate to Box Classifier.
The Box Classification values for your enterprise were created in section 2.3 and may differ from those
values available for Classifier Labels.
The Box Classifier product allows you to use rules to:
 Provide mappings from Classifier Labels to the metadata system used by the Box Cloud Storage
system for classification and access control.
 Prevent documents with specific Classifier labels being retained in the Box Cloud Storage system.
 Provide Adjudicator access within your Box enterprise to a document with a specific Classifier label.
This process is termed “Quarantine”. An Adjudicator is configured as described in Section 2.5.
For example, using the example Box Classification values shown in section 2.3 a set of rules could be:

Figure 3: Example Policy Rules

2.7 The Box Classifier Service


The Box Classifier Service is started from the standard Microsoft Management Control for Services as shown
in Figure 4.

[Link] 9
Box Classifier Guide UM643301

Figure 4: Box Classifier Service

Before using the Box Classifier Service, you should ensure that the Recovery options for the service are set
appropriately. It is recommended that the Restart the Service option is used.

3 FUNCTIONALITY
3.1 Events and Actions
Various events are recorded by the Box Cloud Storage system which can then be queried by Box Classifier.
The events identify the user and the document in question.
Box Classifier queries the Box Cloud Storage system for the following events:
 Uploading a new document to [Link].
 Uploading a new version of an existing document to [Link]. Synchronisation actions may also
give rise to such events.
 Adjustment to the Classification in [Link].
 Copying a document in [Link].
 Unlocking a document in [Link].
 Restoring a document from Trash in [Link].
[Link] 10
Box Classifier Guide UM643301

In each of these instances, Box Classifier may perform actions according to any of the following Box Classifier
Rules that have been defined through Classifier Administration:
 Check that the classification applied to the document in [Link] is appropriate to the Classifier
Label. If the Box classification isn’t appropriate, then it can be adjusted and a comment applied
to the document recording the change.
 Based on a specific Classifier Label value or if a Classifier Label is not present, a rule may be
created to Delete the document from [Link].
 Based on a specific Classifier Label value or if a Classifier Label is not present, a rule may be
created to Quarantine the document in [Link]. See Quarantining documents for more
information on the Quarantine functionality.
Note: If there is no Classifier Label but there is a Classification value applied to the document, a rule
to Quarantine the document will not be actioned. This rule exception is to allow documents without
Classifier Labels to exist in the Box Cloud Storage system and yet still be subject to the access control
facilities provided by the Classification system.

3.2 Uploading a document to [Link]


When a document is uploaded to the Box Cloud Storage system, an UPLOAD event is recorded. The UPLOAD
event can take a short time, typically between 30 seconds and 90 seconds, before it is available to be
queried by Box Classifier. A user might apply a Box classification during this time for example.
The actions described above can be performed on the document.

3.3 Uploading a new version of a document to [Link]


A new version of a document in [Link] will retain the classification value from a previous document
version. The new version may, however, use a different Classifier Label. In this situation, Box Classifier may
need to perform new actions on the revised document.

3.4 Synchronisation of documents


Synchronisation of documents between a user’s desktop or mobile device and [Link] can introduce a new
version similar to Uploading a new version. The actions described above can be performed on the revised
document:

3.5 Adjusting the Classification of a document in [Link]


When a document classification is created, deleted or changed in the Box Cloud Storage system, an event is
recorded. The actions described above can be performed on the revised document:

3.6 Unlocking a document in [Link]


Downloading a locked document from [Link] is usually not possible. For Box Classifier to function
correctly, it must be able to download documents to check the associated Classifier Label. For this reason,
Box Classifier monitors events that record when a document is unlocked.
The actions described above can be performed on the revised document.

[Link] 11
Box Classifier Guide UM643301

3.7 Copying a document in [Link]


When a document is copied in the Box Cloud Storage system, the classification value is not copied from the
original version of the document. Box Classifier monitors the COPY event and will perform the actions
described above on the document copy.

3.8 Restoring a document from Trash in [Link]


When a document is restored in the Box Cloud Storage system, Box Classifier checks the associated Classifier
Label on the document.
The actions described above can be performed on the restored document.

3.9 Preventing documents with specific Classifier labels being retained


in [Link]
Based on a specific Classifier Label value or if a Classifier Label is not present, a rule may be created to Delete
the document from [Link].
If this action is performed, then the document is trashed and then deleted permanently from the Trash
folder in Box.

Note: If there are multiple versions of a document, a delete action on the current version will apply to all
versions of the document. This is a Box Cloud Storage system approach.

If desktop synchronization or mobile synchronization is enabled then this may also delete the document from
those paired machines and/or devices.

3.10 Quarantining documents with specific Classifier labels in [Link]


Box Classifier requires the nomination of an Adjudicator to handle quarantined documents. The adjudicator
is an enterprise user that is qualified to make corrections to documents to ensure they adhere to the
company’s labelling policies. Typical document corrections are:
 Applying a Classifier Label.
 Applying a Classification
 Collaboration with the document owner to ensure policy compliance.
If a document in [Link] is to be quarantined, then Box Classifier will check if the user performing the
action on the document has a quarantine folder structure created. If there is no quarantine folder structure
created, then Box Classifier will create this folder structure in the Adjudicator’s storage area

The two sub-folders Quarantined by Classifier and Released from Quarantine are shared with two and only
two collaborators: the quarantined document owner and the Adjudicator. Access permissions are applied to
the Quarantined by Classifier sub-folder to allow only the Adjudicator to edit documents whereas the
Released from Quarantine sub-folder allows full permissions for the document owner.

[Link] 12
Box Classifier Guide UM643301

Once the quarantine folder structure is in place a document is moved into quarantine by:
1. Checking to see if a document with that name is already in quarantine. If it is that a new name is
created for the document by appending a number in brackets to the document name. For example,
work [Link]  work proposal(1).docx

2. Moving the document into the Quarantined by Classifier folder.


3. Applying a comment to the document to describe what has happened
4. Locking the file
Once a document has been quarantined the Adjudicator can interrogate and edit the document within the
Quarantine area. The Adjudicator can release the document from Quarantine by moving it to the Released
from Quarantine sub-folder. Adjudicators should be careful to ensure the released document has an
appropriate Classification, otherwise Box Classifier may quarantine the document for a second time.

3.10.1 Changing an Adjudicator

If a decision is made to adjust the Adjudicator then both the old adjudicator and new adjudicator will have
co-ownership on the Quarantine folders. If the old adjudicator no longer requires access then that person
can give ownership of the top level quarantine folder for a user to the new adjudicator and then remove
their own collaboration on the folder. The top level quarantine folder for a user has a name similar to

3.11 Using Box Notes in [Link]


Box Classifier does not monitor Box Notes in [Link]

[Link] 13
Box Classifier Guide UM643301

4 SUPPORTING USERS WITH LANGUAGES OTHER


THAN ENGLISH
Box Classifier adds comments to documents within the Box Cloud Storage System whenever it automatically
adjusts classification values or quarantines a document. The comments are, by default, in English.
Box Classifier can write the comments in the configured language for the Box User who edited the
document. A Box User can change their language settings in the Account Settings section of [Link]. Box
Classifier will use this setting along with information held in the [Link] file that is located in
the installation folder for the product.
There are five comments that may be applied to Box Documents. These are described by the following
section in the [Link] file:

<LanguageSettings>

<BoxComments>

<add index="1" ref="QuarantineString1" value="Box Classifier: the


item's document label requires the item be automatically quarantined."
languageCode="en" />

<add index="2" ref="QuarantineString2" value="Box Classifier: there is no


document label for this item. The item has been automatically quarantined."
languageCode="en" />

<add index="3" ref="QuarantineString3" value="The item has been renamed to


'{0}' because of a name conflict in the folder." languageCode="en" />

<add index="4" ref="LabelString1" value="Box Classifier: document label of


{0} requires a classification of {1}. Automatically applying the classification
{1}." languageCode="en" />

<add index="5" ref="LabelString2" value="Box Classifier: document label of


{0} does not match the classification of {1}. Automatically adjusting the
classification to {2}." languageCode="en" />

<add index="6" ref="QuarantineString1" value="Box Classifier: Das


Dokument-Etikett des Elements erfordert, dass das Element automatisch unter
Quarantäne gestellt wird." languageCode="de" />

<add index="7" ref="QuarantineString2" value="Box Classifier: Für dieses


Element ist kein Dokumentetikett vorhanden. Das Element wurde automatisch unter
Quarantäne gestellt." languageCode="de" />

<add index="8" ref="QuarantineString3" value="Das Element wurde aufgrund


eines Namenskonflikts in den Ordner '{0}' umbenannt." languageCode="de" />

<add index="9" ref="LabelString1" value="Box Classifier: Dokument-Label


von {0} erfordert eine Klassifizierung von {1}. Automatisches Anwenden der
Klassifizierung {1}." languageCode="de" />

<add index="10" ref="LabelString2" value="Box Classifier: Dokument-Label


von {0} stimmt nicht mit der Klassifizierung von {1} überein. Automatische
Anpassung der Klassifizierung auf {2}." languageCode="de" />

[Link] 14
Box Classifier Guide UM643301

</BoxComments>

</LanguageSettings>

The section above shows the five comments firstly in English and then in German. An administrator can
choose to add further language variants so long as the following rules are adhered to:
1. Each XML <add/> node must have a unique index attribute.
2. The ref attribute must contain one of the values shown above. Box Classifier uses this attribute to
understand the context of the comment.
3. The value attribute must retain any placeholders of {0}, {1} and {2}. These placeholders are
replaced with appropriate text when the comment is applied.
4. The languageCode attribute is the two character country/region code taken from ISO 639-1
Language Code
Once the file has been edited to include additional languages, restart the Box Classifier Service.

5 LOAD BALANCING
Depending on the size of your Box Enterprise, it may be necessary to install Box Classifier onto multiple
servers to provide load balancing. If you determine this is necessary, then simply adjust the load balancing
filter using the Box Configuration Console.
Box Classifier will use the Filter for the Users Handled field in the configuration form to only handle events
from users that match the Filter. For example, if you have two servers with Box Classifier installed then you
can craft a filter to handle half the users on the first server and use another filter to handle the other half of
the users on the second server.
The filter takes the form of a regular expression. To check that you have crafted the regular expression
correctly or to receive a suggestion click the Adjust… button. The following dialog is presented.

[Link] 15
Box Classifier Guide UM643301

Figure 5: Load Balancing Filter

You can enter a regular expression in the Filter field and click Test to see which users the filter applies to. If
you would like a suggested filter, click the Suggest… button to present the following dialog:

Figure 6: Suggest a Filter

Simply adjust the Number of Servers with Box Classifier installed and This Server’s Number to suggest a
filter. The filter suggested works on the first letter of Users’ Surnames. The example shown in Figure 8,
describes a filter that will handle surnames that begin with a letter in the first half of the alphabet, i.e. A to
M.

6 AUDIT EVENTS
Important events are recorded in the Windows Application Event log under the log name Boldon
James/Classifier. Events have the following general appearance

[Link] 16
Box Classifier Guide UM643301

Figure 7: Example Audit Event

The following important events are recorded:

Level Event Data Meaning

(Event ID)

Error Licensing messages There is a problem with the


licensing for Box Classifier.
(1005)

Information Service started Box Classifier Service started

(2000)

Information Service stopping Box Classifier Service stopping

(2000)

Error The service encountered an issue that it Box Classifier Service stopping
cannot recover from and will be stopped.
(2000) Further diagnostic information…

Information Item Classification has been added to Box Classification added


[Link].
(7000)

Item details and event details are also


recorded

Information Item Classification has been updated in [Link]. Box Classification updated
Item details and event details are also recorded
(7001)

[Link] 17
Box Classifier Guide UM643301

Information Item has been deleted from [Link]. Box Item deleted
Item details are also recorded
(7002)

Information Item has been quarantined in [Link]. Box Item quarantined


Item details are also recorded
(7003)

Warning Quarantine of the item failed in [Link]. Box Item quarantine failed
Item details are also recorded
(7006)

Warning Deletion of the item failed in [Link]. Box Item deletion failed
Item details are also recorded
(7007)

Warning A required change to the item Classification failed Box Classification change failed
in [Link].
(7008)
Item details are also recorded

7 INTERWORKING WITH CLASSIFIER APPLICATIONS


Key information on the interworking with Classifier applications is:
 Box Classifier only interrogates the source item to obtain the Classifier Label (if any) – it
does not modify the original item.
 Box Classifier uses the same Label information as File Classifier, Office Classifier and CAD
Classifier (Classifier ‘private’ label property and Label Locations) when reading labels from
file types supported (e.g. Microsoft Word, Excel, PowerPoint and Project - note VISIO files
must be labelled using the VISIO application).
Alternate Data Stream Limitation:
 Uploading a file labelled by Power or File Classifier (but not supported by Office Classifier)
to the Box Cloud Storage System will lose the additional Alternate Data Stream (ADS)
information. The file will then be unlabelled.

Note: More detailed information on the use of labels by, and propagation of labels between, Classifier
applications can be found in Administration Guide > Appendix – Label Propagation Mechanisms.

[Link] 18
Box Classifier Guide UM643301

8 DEPLOYMENT AND INFRASTRUCTURE


CONSIDERATIONS
8.1 Firewalls and Proxies
Box Classifier connects to your Box Enterprise through the two endpoints of [Link] and
[Link]. You may need to adjust Firewall configuration and Proxy Server configuration to allow this
connection.
Both the Box Classifier Service and Box Classifier Configuration Console make connections to the Box
Enterprise. The Box Classifier Configuration Console connects under the current user’s credentials whereas
the Box Classifier Service runs as Local System account by default. You should change the Box Classifier
Service to run as an appropriate user that is recognised by the Proxy Server.
The Box Classifier Service is started from the standard Microsoft Management Control for Services as shown
in Figure 4.

8.2 Load and Performance


When deploying Box Classifier it is vital to understand the load that will be placed on the product. This
depends on a number of factors including:
 The typical number of document uploads to [Link] and document changes in [Link] per day
 The typical size of documents that are uploaded or changed.
Details on these two factors can be obtained through the Reports facility that [Link] offers on the Box
Admin Console.
Boldon James Box Classifier has the following performance characteristics in order of influence:
1. [Link] API Limits – [Link] impose usage restrictions on 3rd Party integrations. These standard
enterprise restrictions impose an approximate handling limit for Box Classifier of:
 51000 document uploads and document changes per day of a 10KB file
 6900 document uploads and document changes per day of a 250MB file
[Link] can agree to a relaxation of these usage restrictions for an Enterprise. Interested customers
should contact [Link] for further information.
2. Network Bandwidth – the Box Classifier service downloads files temporarily to memory to analyse
the Classifier label of the file. It does not upload files. Network bandwidth plays a more significant
role as document size increases. Testing indicates that a download speed of around 300 Mbps does
not constrain performance for large files of 250MB; it is the API limits imposed by [Link] that
constrain performance when such bandwidth is available. For smaller files of around 10KB then the
download speed can be as low as 20 Mbps before [Link] API limits influence performance.
3. Memory – the Box Classifier service streams files directly to memory to check the associated
Classifier label. This means that peak memory usage occurs when there are multiple large files being
streamed simultaneously. Box Classifier uses ten channels to stream files and allows a simple
heuristic of 10 x <maximum file size> to indicate memory required by the service. For instance,
system testing shows peak private byte usage of 2.4GB when handling files of size 250MB.
[Link] 19
Box Classifier Guide UM643301

Boldon James recommend that Box Classifier is installed on a standalone machine that does not
share resources with other server applications such as Microsoft Exchange.
4. CPU - the Box Classifier service has an average processor utilisation of 0.37% when handling files of
size 10 KB rising to 8.6% when handling files of size 250MB. This was determined using Intel i7-4770K
CPU@3.5GHz
5. Disk – the Box Classifier service streams files directly to memory so disk size and disk performance is
not a primary consideration.
The primary factor that affects the performance of Box Classifier is the [Link] API limits with a secondary
influence of download speed. When Box Classifier approaches the [Link] API limits, user events that occur
within [Link] are queued to be retried at a later time. From a user’s perspective, a document will take
longer to be processed. During peak usage this may be acceptable but if it occurs on a regular basis
customers should consider deploying an additional instance of Box Classifier for load balancing as described
in section 5.

[Link] 20

You might also like