WEBVTT

00:00:10.960 --> 00:00:16.490
Jaql is a loosely typed functional language,
with lazy evaluation. It's part of BigInsights,

00:00:16.490 --> 00:00:20.200
IBM's Hadoop based Big Data platform.

00:00:20.200 --> 00:00:25.259
The initial inspiration for Jaql was a JSON
Query Language that provided query processing

00:00:25.259 --> 00:00:30.470
for semi-structured data, particularly nested
data represented using the Jaql Data Model

00:00:30.470 --> 00:00:32.829
or JDM

00:00:32.829 --> 00:00:38.559
The JDM is a superset of JSON, a standard
data model, that is similar to XML but much

00:00:38.559 --> 00:00:40.649
easier to work with.

00:00:40.649 --> 00:00:47.039
Jaql provides a complete solution environment
and integration point for all the data storage,

00:00:47.039 --> 00:00:53.910
functions and capabilities of BigInsights;
including flexible access to data and analytics.

00:00:53.910 --> 00:00:59.510
Jaql is the functional language that glues
it all together.

00:00:59.510 --> 00:01:05.800
It also provides ways to reach out to external
data and services such as relational databases,

00:01:05.800 --> 00:01:10.909
indexing services, text analytics, machine
learning and so on.

00:01:10.909 --> 00:01:16.200
Jaql provides an elegant programming model
for developing complex and reusable queries

00:01:16.200 --> 00:01:22.460
and transformations against JSON data structures;
even very deeply nested data structures.

00:01:22.460 --> 00:01:28.799
As with SQL, Jaql is a declarative query language
so you focus on what you want to achieve,

00:01:28.799 --> 00:01:31.469
not how it's done.

00:01:31.469 --> 00:01:37.829
Under the hood, Jaql automatically exploits
the massive parallelism of MapReduce. You

00:01:37.829 --> 00:01:42.380
don't have to worry about the complexities
of MapReduce programming.

00:01:42.380 --> 00:01:48.670
Jaql is an extensible language. It provides
native functions and modules, that allow you

00:01:48.670 --> 00:01:54.500
to build up packages of re-usable logic that
can be shared throughout your environment.

00:01:54.500 --> 00:01:59.810
So, when would you choose Jaql instead of
another language?

00:01:59.810 --> 00:02:06.810
1. Jaql is a good choice when your data is
deeply nested, representing complex relationships.

00:02:08.258 --> 00:02:14.840
Jaql can work with arbitrarily nested data
and can even handle data that varies in structure

00:02:14.840 --> 00:02:17.370
and content from record to record.

00:02:17.370 --> 00:02:23.220
2. Jaql is a good choice when you need to
combine programming models because it offers

00:02:23.220 --> 00:02:30.150
both the core Jaql syntax and you can embed
SQL within your Jaql – that's like combining

00:02:30.150 --> 00:02:32.710
Hive and Pig together into one language

00:02:32.710 --> 00:02:39.710
3. Jaql is a good choice when you need a modular
language that lets you write native functions.

00:02:40.270 --> 00:02:46.410
You can parameterize and encapsulate any functionality
of the language. and then package those functions

00:02:46.410 --> 00:02:51.100
into reusable modules to be shared across
all of your applications.

00:02:51.100 --> 00:02:58.100
4. Jaql is also a good choice when you need
a full programming language with both flow-control

00:02:58.680 --> 00:03:04.320
and native functions and modules. You can
develop full programs from the ground up using

00:03:04.320 --> 00:03:09.410
JAQL without requiring the use of another
programming language to orchestrate your logic,

00:03:09.410 --> 00:03:16.410
which is particularly important for data flows.