Getting Started

General XINA information and concepts.

Introduction and Ecosystem
Terms and Concepts
Importing Data
Data Types
Sandbox Quick Start Guide

Introduction and Ecosystem

XINA is an integrated data management platform, developed at NASA GSFC. XINA is provided as a managed service hosted on Amazon Web Services.

Overview

The XINA platform provides five primary functions:

Structured Database Storage (MySQL on AWS RDS)
File Storage (AWS S3)
Task Management (XINA Run on AWS EC2, XINA Lambda on AWS Lambda)
Web Client (XINA Web, Angular)
Direct API Access (XINA Tunnel)

XINA supports Launchpad and NAMS integration for user management and authentication.

Limitations

XINA is not a single standalone application, and cannot currently be installed locally. We provide the software as a service through a NASA AWS account integration. Specific AWS requirements and costs will vary depending on project parameters.

XINA is not recommended as a gold copy for data storage. Although AWS cloud services are highly reliable and XINA is often used as a valuable backup tool, a full gold copy should always be kept onsite.

XINA System Components

The XINA platform is composed of several interconnected components and applications:

XINA Server

The XINA server is the core application of the XINA platform. It manages all incoming and outgoing XINA data and provides API access. The server is built on a MySQL database backend and uses the AWS S3 service for large file storage.

XINA Web

XINA Web (formerly XINA Online) is the primary XINA front-end application, written in TypeScript with Google's Angular web application platform. Authentication is integrated with NASA Launchpad and managed through NAMS.

XINA Tunnel

The XINA Tunnel utility is a Java application intended to facilitate communication with the XINA API. The tunnel connects directly to the XINA server and manages connection security and authentication. It then opens a local webserver to which client applications can connect and communicate with the core XINA server. Full reference is available here.

XINA Import

The XINA Import utility is a Java application to simplify importing data to the XINA server by importing XINA API actions from JSON files. Full reference is available here.

XINA Run

XINA Run is a Java application for managing and executing asynchronous tasks though the XINA platform. Full reference is available here.

XINA Lambda

XINA Lambda is an integrated service for executing asynchronous tasks from the AWS Lambda platform. Full reference is available here.

Terms and Concepts

Database

Databases are the core data storage structures in XINA. A database essentially defines a MySQL table, with additional features managed by the XINA server system.

Each database is defined by a set of fields, which specify the columns of the table. Fields are primarily defined by:

Name, unique to the field in the database
Static data type
Whether a value must be provided by each record (an empty value being null)

A single unit of data in a database is a record, corresponding to a row of the table. Each record contains a value for each field of the database.

Structural database changes (adding / changing / removing fields) are very slow (hours to days for very large databases) so initial time investment to optimize database requirements is worthwhile.

Group

Databases in XINA are organized into a heirarchical structure of groups, which can each contain any number of groups and databases. For example:

The model group contains a journal database and data group
The data group contains a housekeeping database and science database

A dot notation is used to reference groups and databases. For example, moma.data.science refers to the science database in the above configuration.

Importing Data

There are several approaches for importing data into XINA, but for most projects we recommend the XINA Import utility.

XINA Import reads XINA API calls from JSON files and passes them to the XINA server. Each JSON file corresponds to a single API action, but may be paired with additional files of other types depending on the content.

For example, to upload some housekeeping data from a CSV requires two files. First, the JSON file:

{
  "action": "load",
  "database": "demo.model.data.hk.full",
  "columns": true,
  "delimit": ",",
  "line": "\n",
  "$object_id": "{local}/hk.csv"
}

The CSV file then looks like:

t,name,value
1602086313288000,SCAN_INDEX(Step),-1
1602086313288000,MO1_LD1_CURR(mA),0
1602086313288000,MO1_LD2_CURR(mA),0
1602086313288000,MO1_CASE_TEC(C),21.739
...

A couple notes on these:

The "line" property in the JSON file must exactly match the new line character(s) in the CSV file. You can use "\r\n" if you prefer, but either way we recommend explicitly using one or the other when you write the files, as a general "print line" may use different output depending on the platform.
The "{local}" in the "$object_id" property is a macro used by the XINA Import application to look for the CSV file in the same location as the JSON file. We recommend using this and keeping the files in the same location. If you need to separate them you can use a full path instead, but this is more fragile if folder organization needs to be changed.
For best performance we recommend paging the CSV files so that each is 50MB or less.
If you need to represent an empty value, you can either omit a data point or use "NULL" (without quotes). ("NaN" and "Infinity" are not supported at the database level.)

While this CSV approach is recommended for large data sets, data can alternatively be embedded directly in JSON files. For example, a file to insert a new instant might look like:

{
  "action": "insert",
  "database": "demo.model.data.ins",
  "records": [
    {
      "u_id": "58ea870a-52c3-33c7-b858-c20795ec3301",
      "p_id": 0,
      "s_id": 0,
      "type": 20,
      "level": 0,
      "t": 1606333792000000,
      "label": "SPECTRA_Startf-0_Stopf-1k",
      "content": "some additional text here...",
      "meta": {
        "Resolution Bandwidth": 2.07014,
        "Stop Frequency": 1000,
        "Average Factor": 30,
        "Start Frequency": 0
      }
    }
  ]
}

The full API reference can be found here.

Data Types

XINA has a fixed set of data types which apply to attributes and fields. They are intended to provide consistent behavior across MySQL, Java, and JavaScript data types.

Numeric Types

Type	Java	MySQL	JavaScript	Notes
`int(1)`	`byte`	`tinyint`	`number`	integer, -2⁷ to 2⁷-1
`int(2)`	`short`	`smallint`	`number`	integer, -2¹⁵ to 2¹⁵-1
`int(4)`	`int`	`int`	`number`	integer, -2³¹ to 2³¹-1
`int(8)`	`long`	`bigint`	`number`	integer, -2⁶³ to 2⁶³-1 ⚠️
`float(4)`	`float`	`float`	`number`	IEEE 754 4 byte floating point
`float(8)`	`double`	`double`	`number`	IEEE 754 8 byte floating point
`boolean`	`boolean`	`tinyint`	`boolean`	MySQL treats 0 as `false`, non-zero as `true`

⚠️ JavaScript number is 8 byte float, so only -2⁵³ to 2⁵³-1 is stored with exact precision

Character Types

Character data types offer two encoding options:

UTF-8 - default encoding, variable length, 1 to 4 bytes per character
ASCII - subset of UTF-8, fixed length, 1 byte per character

Two SQL types:

char(n) - data stored in the table, fastest search and index, uses fixed amount of space per row (n * max_bytes_per_character)
varchar(n) - data stored in the table, fast search and index, uses variable amount of space per row (up to n * max_bytes_per_character)
text - data stored outside the table, slower search and index, uses only as much space as needed

Two general types:

string - text is normalized before insertion
- leading and trailing whitespace is trimmed
- all internal whitespace is reduced to a single space character
text - text is inserted as provided

Note, all string operations are case-insensitive by default. This can be overridden with the collate expression by specifying a binary collation.

Type	Java	MySQL	JavaScript	Notes
`utf8string(n)`	`string`	`char(n)`	`string`	n up to 128, uses n*4 bytes, normalized
`utf8vstring(n)`	`string`	`varchar(n)`	`string`	n up to 128, uses up to n*4 bytes, normalized
`utf8string`	`string`	`mediumtext`	`string`	up to 2²⁴ bytes, normalized
`utf8text`	`string`	`mediumtext`	`string`	up to 2²⁴ bytes, not normalized
`asciistring(n)`	`string`	`char(n)`	`string`	n up to 256, uses n bytes, normalized
`asciivstring(n)`	`string`	`varchar(n)`	`string`	n up to 256, uses up to n bytes, normalized
`asciistring`	`string`	`mediumtext`	`string`	up to 2²⁴ bytes, normalized
`asciitext`	`string`	`mediumtext`	`string`	up to 2²⁴ bytes, not normalized

Temporal Types

Temporal data types store time data. There are two categories of temporal types:

instants - identify specific moment in time, independent of time zone
- stored numerically in the database in milliseconds
- datetime and date use Unix epoch
- datetime and date comparable in database
- date + time = datetime
- typically displayed in local time zone in front-end applications
timestamps - identify specific formatted time without time zone consideration (thus local)
- stored as ISO 8601 formatted string in database
- localdate and localdatetime comparable in database
- CONCAT(localdate, 'T', localtime) = localdatetime

Type	Java	MySQL	JavaScript	Notes
`datetime`	`DateTime`	`bigint`	`date`	instant with millisecond precision, as Unix time
`date`	`XDate`	`bigint`	`date`	instant at start of date UTC, as Unix time
`time`	`LocalTime`	`int`	`number`	length of time up to 23:59:59.999, as millisecond count
`localdatetime`	`LocalDateTime`	`char(24)`	`string`	full timestamp without timezone, stored as `string`
`localdate`	`LocalDate`	`char(10)`	`string`	date without timezone, stored as `string`
`localtime`	`LocalTime`	`char(12)`	`string`	length of time up to 23:59:59.999, as `string`

JSON Types

JSON data types store JSON data directly in the database.

Type	Java	MySQL	JavaScript
`json`	`JsonValue`	`json`	`*`
`jsonarray`	`JsonArray`	`json`	`array`
`jsonobject`	`JsonObject`	`json`	`object`

Sandbox Quick Start Guide

In order to streamline onboarding for new XINA projects we have created a XINA "Sandbox" environment to test data pipelines and tools. Everything in the sandbox is fully configured as a typical XINA production environment.

Step 1: Request NAMS Access

To get started, first request access through the NAMS service. The application name is "GSFC XINA Sandbox".

Once you receive confirmation that the account is approved, perform your first login to the XINA Sandbox by going to sandbox.xina.io. This initial login creates your user account. You will initially have access to a series of default Sandbox data. If you require access for a specific project contact our team and we will help finish setting up required permissions.

Step 3: Create an API Key

To access the XINA API for importing data you will require a XINA API key. In the XINA web application, click your user name in the top right, then "User Profile" in the drop down.

Switch to the "Access Keys" tab and click "Create New Key".

Copy the entire displayed text and save it to a local file. The key will only be displayed once upon creation, if lost you will need to make a new key.

Step 4: Download XINA Tunnel and XINA Import

The XINA Tunnel and XINA Import utilities are the recommended starting point for importing data. Details for each are available on their respective wiki pages.

Step 5: Import Sample Data

UNDER CONSTRUCTION