Microsoft Azure Data Lake Storage (ADLS) Gen2 File System - Import

Bridge Specifications

Vendor	Microsoft
Tool Name	Azure Data Lake Storage (ADLS) File System
Tool Version	Gen2
Tool Web Site	https://azure.microsoft.com/en-us/services/storage/data-lake-storage/
Supported Methodology	[File System] Multi-Model, Data Store (NoSQL / Hierarchical) via Java API
Data Profiling
Incremental Harvesting
Multi-Model Harvesting
Remote Repository Browsing for Model Selection

SPECIFICATIONS
Tool: Microsoft / Azure Data Lake Storage (ADLS) File System version Gen2 via Java API
See https://azure.microsoft.com/en-us/services/storage/data-lake-storage/
Metadata: [File System] Multi-Model, Data Store (NoSQL / Hierarchical)
Component: MicrosoftAzureDataLakeStorageGen2 version 11.2.0

DISCLAIMER
This import bridge requires internet access to download third-party libraries:
- such as https://repo.maven.apache.org/maven2/ to download open source third-party libraries,
- and more sites for other third-party software such as database specific JDBC drivers.

The downloaded third-party libraries are stored into $HOME/data/download/MIMB/
- If HTTPS fails, the import bridge then tries with HTTP.
- If a proxy is used to access internet, you must configure that proxy in the JRE (see the -j option in the Miscellaneous parameter).
- If the import bridge does not have full access to internet, that $HOME/data/download/MIMB/ directory can be copied from another server with internet access where the command $HOME/bin/MIMB.sh (or .bat) -d can be used to download all third-party libraries used by all bridges at once.

By running this import bridge, you hereby acknowledge responsibility for the license terms and any potential security vulnerabilities from these downloaded third-party software libraries.

OVERVIEW
This import bridge crawls a data lake implemented on the Microsoft Azure Data Lake Storage Service to detect (reverse engineer) metadata from all the files (for data catalog purpose).
This includes data sampling driven metadata discovery of the data structure (e.g. CSV table, JSON hierarchy) and data types (e.g. Integer, Date, String).

REQUIREMENTS
n/a

AUTHENTICATION
Use one of the following authentication methods based on the Account and Secret parameter values:

- Microsoft Azure access key:
Account = Storage Account
Secret = Access Key, e.g.
CHooPyc1B2ec4WAYcC8o/Yw6kxEyXGx6QhOXw7sABH2GbK6bpgJb7BT73dneQiW9jlQzxDssbkWycSMIeZIVfA=

- Microsoft Azure Shared Access Signature (SAS) token:
Account = Storage Account
Secret = SAS token, e.g.
?sv=2015-12-11&ss=bfqt&srt=sco&sp=rwdlazxpx&se=2040-10-01T19:16:45Z&st=2020-10-01T11:16:35Z&spr=https&sig=PUDjdH3wZ2NqO49pNJIex5sIz9OKMBSE77b4pbKWXxI%3D
For more information, see https://docs.microsoft.com/en-us/azure/storage/common/storage-sas-overview

- Microsoft Azure Key Vault secret:
Account = Storage account
Secret = VaultURL, e.g.
For Azure: https://My-vault-01.vault.azure.net/secrets/My-secret-01
For Amazon: arn:aws:secretsmanager:us-east-1:111222333444:secret:My-secret-01"
For Google: projects/351268939776/secrets/My-secret-01/versions/1
For more information, see https://docs.microsoft.com/en-us/azure/key-vault/secrets/

- Microsoft Azure Temporary credentials (Externally MM managed cloud identity):
Account = Storage account
Secret = ""
This is the first authorization method tried by default when Secret parameter is empty.
For more information, see your Metadata Management (MM) / Data Catalog / Data Governance application documentation on managing cloud identity for metadata harvesting.

- Microsoft Azure Managed Service Identity (MSI):
Account = Storage account
Secret = /
or
Secret = TenantID/Application(client) ID, e.g
4d33abfb-7bd8-54ba-bc97-5890aaa16040/ba8f6a23-3d6e-4fbf-a561-2e36a79c3701
A VM can be configured to have multiple user identities and you should specify the one you want to use with its TenantID/ApplicationID.
When VM has one identity (e.g., system) you do not need to know and specify its details but the '/' character, like Secret = /.
For more information, see https://docs.microsoft.com/en-us/azure/active-directory/managed-identities-azure-resources, https://hadoop.apache.org/docs/stable/hadoop-azure/abfs.html#Azure_Managed_Identity

- Azure Data Lake Storage database:
The user should have the following permission "Storage Blob Data Reader" to perform Import, Test and Browse actions.

- Active Directory:
In order to use 'Active Directory authentication', add a new 'Cloud Identity' entity under Manage Cloud Identities. Specify the required attributes:

Type - Microsoft Azure
Directory (TENANT) ID
Application (CLIENT) ID
Application (CLIENT) Secret

This 'Cloud Identity' should be specified under the import setup of this bridge.
The 'Secret' parameter should be empty if the 'Active Directory authentication' method is utilized.

PERMISSIONS
n/a

FREQUENTLY ASKED QUESTIONS
Q: For my virtual machine, what should I replace for the "System assigned managed identity"?
A: You should replace the Client-ID with an Application-ID from the automatically created enterprise application.

Q: What is the purpose of having the include filter, if I specified the partition directories?
A: We use the include filter for faster import and to not scan all folders under provided roots. So, it is not necessary to specify include filter parameter to recognize partitions.

Q: How can I instruct the bridge to treat all directories under the provided root(s) as partitions with one-level hierarchy and to use its subdirectory names as partitioned feature names?
A: [*]/{*}/*.csv

Q: How may I connect to ADLS that is configured for Private Blob Storage?
A: Private Blob storage indicates that public access to the storage account is disabled and a private endpoint connection was established (private link). This endpoint permits access via a specific virtual network to access this Blob Storage account. Due to the security implemented via Private Blob Storage, a Private link with DFS target sub-resource as well as a Private link with BLOB target sub-resource should be created. Since the bridge enumerates and check containers via the BLOB API, this API is involved in the processing of metadata. In the end, the two private links need to exist: the DFS target sub-resource and the second private link with BLOB target sub-resource, as well as the aforementioned 'Role" referenced under the "Authentication" section of this bridge document.
Lastly, the hosts file must be edited on the MM Server Operating System - as further explained below.
In order to successfully connect to the configured Private Blob Storage account, you must modify the MM Server Operating System hosts file to include the hostname of your Blob storage and the Privatelink URL, e.g.
*MM IP* abc123.blob.core.windows.net
*MM IP* abc123.privatelink.blob.core.windows.net

Additionally, the 'Root directory' parameter must be configured to connect via the File System name instead of the "Hierarchical" hostname.

LIMITATIONS
Refer to the current general known limitations at https://metaintegration.com/Products/MIMB/Help/#!Documents/mimbknownlimitations.html
n/a

SUPPORT
Provide a troubleshooting package with:
- the debug log (can be set in the UI or in conf/conf.properties with MIR_LOG_LEVEL=6)
- the metadata backup if available (can be set in the Miscellaneous parameter with -backup option, although this common option is not implemented on all bridges for technical reasons).
n/a

SUPPORTED FILES
Data Definition / Schema / Metadata file formats (no data):
- Fixed Width files typically from mainframe (see details below)
- COBOL COPYBOOK files typically from mainframe (see details below)
- W3C XML XSD (XML Schema Definition)

Text data file formats (data sampling driven metadata discovery):
- Delimited (Flat) files such as CSV (see details below)
- Open Office Excel XML .XSLX (see details below)
- W3C XML (not defined from XML XSD)
- JSON (JavaScript Object Notation) (see details below)

Binary data file formats (which include a schema definition as header or footer):
- Apache Avro (see details below)
- Apache Parquet (see details below)
- Apache ORC (see details below)

as well as the compressed versions of the above formats:
- ZIP (as a compression format, not as archive format)
- BZIP
- GZIP
- LZ4
- Snappy (as standard Snappy format, not as Hadoop native Snappy format)

DELIMITED FILES
This bridge detects (reverse engineer) the metadata from a data file of type Delimited File (also known as Flat File).
The detection of such Delimited File is not based on file extensions (such as .CSV, .PSV) but rather by sampling the file content.

The bridge can detect a header row, and use it to create the field name, otherwise generic field names are created.

The bridge samples up to 100 rows in order to automatically detect the field separators which by default include:
', (comma)', '; (semicolon)', ': (colon)', '\t (tab)', '| (pipe)', '0x1 (ctrl+A)', 'BS (\u0008)'
More separators can be added in the auto detection process (including double characters), see the Miscellaneous parameter.

During the sampling, the bridge also detects the file data types, such as DATE, NUMBER, STRING.

FIXED WIDTH FILES
This bridge creates metadata for data files of type Fixed Width File.
Such metadata cannot be automatically detected (reverse engineered) by sampling the data files (e.g. customers.dat or even just customers with no extension).
Therefore, this bridge imports a 'Fixed Width File Definition' file which must be with extension .fixed_width_file_definition format file
(e.g. customers.dat.fixed_width_file_definition format file will create the metadata of a file named file customers with the fields defined inside)
This is the equivalent of a RDBMS DDL for fixed width files. With such a long extension, this data definition file can coexist with the actual data files in the each file system directory containing them.

The 'Fixed Width File Definition' file format is defined as follows:
- Format file must start with the following header:
column name, offset, width, data type, comment
- All offsets must be unique and greater than or equal to 0.
a,0
b,4
- The file format is invalid when some columns have offsets and others don't.
a,0
b,
c,4
- When all columns do not have offsets but have widths the application assumes that columns are ordered and calculates offsets based on widths.
a,,4 -> a,1,4
b,,25 -> b,5,25
- When the offset is present the application ignores widths as they are calculated from the offsets.
a,1,4
b,5,25
- Types and comments are used as documentation only.
a,1,4,int
b,5,25,char[25],identifier

This bridge detects the following data types: INTEGER, FLOAT, STRING, DATE, BOOLEAN.

COBOL COPYBOOK FILES
This bridge can only import the COBOL COPYBOOK files (which contain the data definitions), therefore does not detect (reverse engineer) metadata from actual COBOL data files.
The detection of such COBOL COPYBOOK file is not based on file extensions (such as .CPY) but rather by sampling the file content.

This bridge creates a 'Physical Hierarchical Model' which reflects a truly flat, byte-position defined, record structure, which is useful for stitching to the DI/ETL processes. Therefore, the physical model has all the physical elements required to define a flat record, which is ONE table with all the elements (including multiple columns for OCCURS elements when the proper bridge parameter is set).

Note that this bridge does not currently support the COPY verb, and reports a parsing error at the line and position at which the COPY statement begins. In order to import Copybooks with the Copy Statement, create an expanded Copybook file with the included sections already in place (replacing the COPY verb). Most COBOL compilers have the option to output only the preprocessed Copybooks with the COPY and REPLACE statements expanded.

Frequently Asked Questions:
Q: Why is the default start column '6' (six) and the default end column '72' (seventy-two)?
A: The bridge parser counts columns starting at 0 (zero), rather than 1 (one). Thus, the defaults leave the standard first six columns for line numbers, next column for comment indicators, and last 8 columns (out of 80) for additional line comment information.

EXCEL (XLSX) FILES
This bridge detects (reverse engineer) the metadata from a data file of type Excel XML format (XLSX).
The detection of such Excel file is based on file extension .XLSX.

The bridge can detect a header row, and use it to create the field name, otherwise generic field names are created.

The bridge samples up to 1000 rows to detect the file data types, such as DATE, NUMBER, STRING.

If an Excel file has multiple sheets, each sheet is imported as the equivalent of a file/table with the same sheet name.

This bridge uses the machine's local file system to read files and allows you to specify the character set encoding files use.

This bridge only imports the CSV aspect of Excel, but does not support any BI / Analysis aspects of Excel like pivot tables, charts, etc.

W3C XML FILES
This W3C XML import bridge is used in conjunction with other file import bridges (e.g. CSV, XLSX, Json, Avro, Parquet) by all data lake / file crawler import bridges (e.g. File systems, Amazon S3, Hadoop HDFS).

The purpose of this XML import is to reverse engineer a model/schema from its content, when such XML was not formally defined by an XML Schema (XSD or DTD).
Such XML files are common from IoT devices uploaded into a data lake.

Nevertheless, such XML files are expected to be fully W3C compliant, especially with respect to the XML text declaration, well-formed parsed entities, and character encoding of entities.
See W3C standards for more details:
https://www.w3.org/TR/xml/#sec-TextDecl

Warning, you must use the dedicated XML based import bridges for all other needs such as:
- other standard W3C XML import bridges (e.g. DTD, XSD, WSDL, OWL/RDL)
- tool specific XML import bridges (e.g. Erwin Data Modeler XML, Informatica PowerCenter XML)

JSON FILES
This bridge imports metadata from JSON files using the Java API.
This bridge loads the entire JSON file using a streaming parser, therefore there are no size limits, although it may take time if it is a remote large JSON file.
This bridge extracts the metadata (JSON hierarchical structure) and detects the following standard JSON data types:
as defined in https://www.json.org/
- String {"stringSample" : "some text", "stringDateSample" : "Thu Apr 06 2017 09:41:51 GMT+0300 (FLE Standard Time)", "expStringSample" : "2.99792458e8"}
- Number {"expNumberSample": 2.99792458E8, "numberSample": 3, "floatSample": 3.141592653589793}
- Array {"arraySample": [1,2,3]}
- True {"booleanSample": true}
- False {"booleanSample": false}
- Null {"nullSample": null}

In addition, the following implementation specific data types are supported:
MongoDB extension:
- The identifier {"_id": {"$oid": "50a9c951300493f64fbffdb6"}}
- Date {"dateExample" : { "$date" : "2014-01-01T05:00:00.000Z"}}
- POSIX date {"isoDateExample" : { "$date" : 1491461103897 }}
- Timestamp {"timestampExample" : { "$timestamp" : { "t" : 1412180887, "i" : 1 } }}
- Number {"numberLongExample": {"$numberLong": "7494814965"}}

CouchDB extension:
- The identifier {"_id":"someId","_rev":"1232343467"}

APACHE AVRO FILES
This bridge imports metadata from Avro files using a Java API.
Note that this bridge is not performing any data driven metadata discovery, but instead reading the schema definition at the header (top) of the Avro file.

This bridge detects the following standard Avro data types:
https://avro.apache.org/docs/current/spec.html#schema_primitive

null - no value.
boolean - a binary value.
int - a 32-bit signed integer.
long - a 64-bit signed integer.
float - a single precision (32 bit) IEEE 754 floating-point number.
double - double precision (64-bit) IEEE 754 floating-point number.
bytes - sequence of 8-bit unsigned bytes.
string - Unicode character sequence.

APACHE PARQUET FILES
This bridge imports metadata from Parquet files using a Java API.
Note that this bridge is not performing any data driven metadata discovery, but instead reading the schema definition at the footer (bottom) of the Parquet file. Therefore, this bridge needs to load the entire Parquet file to reach the schema definition at the end.

If the Parquet file is not compressed, there are no file size limit as the bridge automatically skips the data portion until the footer (although this may take time on large Parquet files). However, if the Parquet file is compressed, then the bridge needs to download the entire file to uncompress it to start with. Therefore, in such case, there is a default file size limit of 10 MB (any bigger files will be ignored), however this limit can be increased in in the Miscellaneous parameter.

This bridge detects the following standard Parquet data types:
as defined in https://parquet.apache.org/documentation/latest

BOOLEAN: 1 bit boolean
INT32: 32 bit signed ints
INT64: 64 bit signed ints
INT96: 96 bit signed ints
FLOAT: IEEE 32-bit floating point values
DOUBLE: IEEE 64-bit floating point values
BYTE_ARRAY: arbitrarily long byte arrays.

APACHE ORC FILES
This bridge imports metadata from ORC files using a Java API.
Note that this bridge is not performing any data driven metadata discovery, but instead reading the schema definition at the header (top) of the ORC file.

This bridge detects the following standard ORC data type:
as defined in https://orc.apache.org/docs/types.html

Integer: boolean (1 bit), tinyint (8 bit), smallint (16 bit), int (32 bit), bigint (64 bit)
Floating point: float, double
String types: string, char, varchar
Binary blobs: binary
Date/time: timestamp, timestamp with local time zone, date
Compound types: struct, list, map, union

MORE INFORMATION
Please refer to the individual parameter's tooltips for more detailed examples.
n/a

Bridge Parameters

Bridge Mapping

Mapping information is not available