Data Storage

Sooner or later you want to retrieve or store some data somewhere. When modelling a pipeline it doesn’t always matter where. Especially when working at high abstraction levels you don’t want to make a choice of data store (just yet). We model a data store after a common symbol for a relational database table in UML, but don’t be fooled, the symbol just represent “a datastore” not a relational datastore specifically.

We mark the type of the datastore by the first letter or a short abbreviation of the actual datastore.

Origin of the Symbol

Modelled after the table and class icons in UML. It’s easy to draw on a whiteboard Start with something that comes close to a square and then at about 1/5th of the top, draw a horizontal line. In software, draw a square first, then position a small rectangle on top and join the shapes. That way you can write the letter of your datastore in the centre of the bottom rectangle.

Data Store Types

Below is a non extensive list of data store types. As with the processors the naming is (roughly) as follows:

• Use the first letter of the data store, capitalized. Unless the datastore has a standard abbreviation of 2 characters.
• If the resulting letter is ambiguous, this is no problem if it’s clear (to you and your team) from the diagram what the data store is.
• To disambiguate add an extra letter, or a ‘.’.
• For Aurora, add the the A after the database type.

Data store	Symbol
DynamoDB	D
Postgresql	P
S3	S3
Redis or Redshift	R
Redis	RD
RedShift	RS
MySQL or Memcached	M
MySQL	M.
Memcached	MD
Microsoft SQL	MS
Aurora	A
Aurora MySQL	MA
Aurora Postgres	PA
Memcached	MC
Oracle	O

Tables

The icon for a datastore is used to mark an individual table or collection. For example, if you indicate you want to store something in MySQL you use a datastore icon for each table you are accessing data from or writing into. Same goes for DynamoDB: you draw a data store for each table. For S3, you draw 1 data store icon per bucket. etc.