Welcome to the 'Intro to Cassandra for Developers' workshop! In this two-hour workshop, the Developer Advocate team of DataStax shows the most important fundamentals and basics of the powerful distributed NoSQL database Apache Cassandra. Using Astra DB, the cloud based Cassandra-as-a-Service platform delivered by DataStax, we will cover the very first steps for every developer who wants to try to learn a new database: creating tables and CRUD operations.
It doesn't matter if you join our workshop live or you prefer to do at your own pace, we have you covered. In this repository, you'll find everything you need for this workshop:
- Materials used during presentations
- Hands-on exercises (see below)
- Workshop video
- Discord chat
- Questions and Answers
To complete the workshop and get a verified badge, follow these simple steps:
- Watch the workshop live or recorded.
- Complete the workshop practice as described below and make the screenshot of the last step (result of the
DELETE
in "Execute CRUD", see here). - Complete the mini-course Cassandra Query Language and take a screenshot of the final screen (the one with buttons "Back"/"Restart" ... + console on the right).
- Complete the mini-course "Cassandra Data Modeling / Digital Library": lessons and practice. Take a screenshot of the final screen of the practice, with the console output at the right.
- Submit the Homework through this form and attach the screenshot(s) you took.
- Give us a few days to review your submission and relax: your well-earned badge will soon land in your mailbox!
Title | Description |
---|---|
Slide deck | Slide deck for the workshop |
1. Create your Astra DB instance | Create your Astra DB instance |
2. Create tables | Create tables |
3. Execute CRUD (Create, Read, Update, Delete) operations | Execute CRUD operations |
ASTRA DB
is the simplest way to run Cassandra with zero operations at all - just push the button and get your cluster. No credit card required, $25.00 USD credit every month, meaning 20M read/write operations and about 80GB storage monthly - sufficient to run small production workloads.
β
Register (if needed) and Sign In to Astra DB https://astra.datastax.com: You can use your Github
, Google
accounts or register with an email
.
Make sure to chose a password with minimum 8 characters, containing upper and lowercase letters, at least one number and special character
β Choose "Start Free Now"
Choose the "Start Free Now" plan, then "Get Started" to work in the free tier.
You will have plenty of free initial credit (renewed each month!), roughly corresponding to 80 GB of storage and 20M read/write operations.
If this is not enough for you, congratulations! You are most likely running a mid- to large-sized business! In that case you should switch to a paid plan.
(You can follow this guide to set up your free-tier database with the $25 monthly credit.)
To create the database, please note that the db_name
and ks_name
in the above image are just placeholders:
-
For the database name - use
workshops
. While Astra DB allows you to fill in these fields with values of your own choosing, please follow our recommendations to ensure the application runs properly. -
For the keyspace name - use
chatsandra
. Please stick to this name, it will make the following steps much easier (you have to customize here and there otherwise). In short:
Note: if you already have a workshops
database, for instance from a previous workshop with us, you can simply create the keyspace with the Add Keyspace
button in your Astra DB dashboard: the new keyspace will be available in few seconds.
Parameter | Value |
---|---|
Database name | workshops |
Keyspace name | chatsandra |
-
For provider and region: Choose any provider (either GCP, AWS or Azure). Region is where your database will reside physically (choose one close to you or your users).
-
Create the database. Review all the fields to make sure they are as shown, and click the
Create Database
button.
You will see your new database as Pending
in the Dashboard;
the status will change to Active
when the database is ready. This will only take 2-3 minutes
(you will also receive an email when it is ready).
β οΈ ImportantThe instructor might show you on screen how to create a token but will have to destroy to token immediately for security reasons.
Ok, now that you have a database created the next step is to create tables to work with.
General Methodology Note: We'll work with a (rather simplified) "chat application" called ChatSandra: users, identified by a unique ID, write posts in several "rooms". Rooms are also uniquely identified by their name, such as
#gardening
. The design of our application is such that we need to be able to (a) retrieve all posts by a given user, sorted by descending date, and (b) retrieve all posts for a given room, sorted by descending date. As dictated by the best practices of data modeling with Cassandra, these requirements are satisfied by creating two very similar tables (denormalization), as you'll see momentarily: they will contain the same posts, but stored (a.k.a. partitioned) in two different ways; and it will be our (that is, the application's) responsibility to maintain them aligned. Of course, we also need ausers
table - we will start with this one indeed.
β Step 2a. Navigate to the CQL Console and login to the database
In the Summary screen for your database, select CQL Console from the top menu in the main window. This will take you to the CQL Console and automatically log you in.
Note: if you are working with your own Cassandra cluster (other than Astra DB), you will reach the CQL Console differently. Moreover, in that case you have to manually create the keyspace once in the CQL Console: this is done with a command similar to
CREATE KEYSPACE chatsandra WITH REPLICATION = {'class': 'NetworkTopologyStrategy', 'replication_factor': 3};
. See the Cassandra documentation for more details on this.
β Step 2b. Describe keyspaces and USE one of them
Ok, now we're ready to rock. Creating tables is quite easy, but before we create one we need to tell the database which keyspace we are working with.
First, let's DESCRIBE all of the keyspaces that are in the database. This will give us a list of the available keyspaces.
π Command to execute
DESC KEYSPACES;
"desc" is short for "describe", either is valid.
CQL commands usually end with a semicolon
;
. If you hit Enter, nothing happens and you don't even get your prompt back, most likely it's because you have not closed the command with;
. If in trouble, you can always get back to the prompt withCtrl-C
and start typing the command anew.
π Expected output
Depending on your setup you might see a different set of keyspaces than in the image. The one we care about for now is chatsandra. From here, execute the USE command with the chatsandra keyspace to tell the database our context is within chatsandra.
Take advantage of the TAB-completion in the CQL Console. Try typing
use cha
and then pressing TAB, for example.
π Command to execute
USE chatsandra;
π Expected output
Notice how the prompt displays <username>@cqlsh:chatsandra>
informing us we are using the chatsandra keyspace. Now we are ready to create our table.
β Step 2c. Create the users table
At this point we can execute a command to create the users table. Just copy/paste the following command into your CQL console at the prompt. Try to identify the primary key, the partition key and the clustering columns (if any) for this table in the command:
π Command to execute
CREATE TABLE IF NOT EXISTS users (
email TEXT,
name TEXT,
password TEXT,
user_id UUID,
PRIMARY KEY (( email ))
);
Then DESCRIBE your keyspace tables to ensure it is there.
π Command to execute
DESC TABLES;
π Expected output
Aaaand BOOM, you created a table in your database. That's it. Now let's go ahead and create a couple more tables before we do something interesting with the data.
β Step 2d. Create the tables for posts
Let us create two more tables, which will contain the posts.
As remarked earlier, we will store the posts in two tables which
differ in how they are partitioned: look at the commands below,
the differences mostly lie in the PRIMARY KEY
specification:
π Command to execute
CREATE TABLE IF NOT EXISTS posts_by_user (
user_id UUID,
post_id TIMEUUID,
room_id TEXT,
text TEXT,
PRIMARY KEY ((user_id), post_id)
) WITH CLUSTERING ORDER BY (post_id DESC);
CREATE TABLE IF NOT EXISTS posts_by_room (
room_id TEXT,
post_id TIMEUUID,
user_id UUID,
text TEXT,
PRIMARY KEY ((room_id), post_id)
) WITH CLUSTERING ORDER BY (post_id DESC);
Then DESCRIBE your keyspace tables: you should see all three listed.
π Command to execute
DESC TABLES;
π Expected output
You may wonder, how did we arrive at this particular structure for the post tables? The answer lies in the methodology for data modeling with Cassandra, which, at its very core, states: first look at the application's needs, determine the required workflows, then map them to a number of queries, finally design a table around each query. We create table posts_by_user to support a query such as "get all posts by a user X"; then we also need table posts_by_room for a query of type "get all posts in room Y". The two tables have the same columns, but the different choice of partition key is what will make the two queries possible on the respective tables.
CRUD stands for "create, read, update, and delete". Simply put, they are the basic types of commands you need to work with ANY database in order to maintain data for your applications.
β Step 3a. (C)RUD = create = insert data, users
Our tables are in place so let's put some data in them. This is done with the INSERT statement. We'll start by inserting three rows into the users table.
Note that we have three users in this example: "111...", "555..." and "999...", which are having some pleasant conversations. In a real application, you would probably generate user IDs at the application level or with the
UUID()
primitive offered by CQL. See the documentation for more details on time/uuid-related CQL functions.
Copy and paste the following in your CQL Console: (Once you have carefully examined the first of the following INSERT statements below, you can simply copy/paste the others which are very similar.)
π Commands to execute
INSERT INTO users (
email, // TEXT
name, // TEXT
password, // TEXT
user_id // UUID: id of a user
)
VALUES (
'[email protected]',
'Otzi Oney',
'123456',
11111111-1111-1111-1111-111111111111
);
INSERT INTO users (email, name, password, user_id) VALUES (
'[email protected]', 'Fred Fivey', 'qwerty',
55555555-5555-5555-5555-555555555555
);
INSERT INTO users (email, name, password, user_id) VALUES (
'[email protected]', 'Nina Niney', 's3cr3t',
99999999-9999-9999-9999-999999999999
);
β Step 3b. (C)RUD = create = insert data, posts
Let's run some more INSERT statements, this time for posts. We'll insert data into the posts_by_user table. (Once you have carefully examined the first of the following INSERT statements below, you can simply copy/paste the others which are very similar.)
Note: in the following, we are using
TIMEUUID
s crafted by hand, to make things easier to visualize. In a real application, you would generate them at application level or, in some cases, using theNOW()
primitive offered by CQL. In the values below, you can just pay attention to the first octet of hex digits.
π Commands to execute
// Insert some data in the "posts_by_user" table
INSERT INTO posts_by_user (
user_id, // UUID: unique id for a user
post_id, // TIMEUUID: unique uuid + timestamp
room_id, // TEXT: id of a chat room
text // TEXT: the post content itself
)
VALUES (
11111111-1111-1111-1111-111111111111,
22222222-5cff-11ec-be16-1fedb0dfd057,
'#hiking',
'I climbed Mt. Gumbo yesterday ...'
);
INSERT INTO posts_by_user (user_id, post_id, room_id, text) VALUES (
11111111-1111-1111-1111-111111111111,
77777777-5cff-11ec-be16-1fedb0dfd057,
'#running', 'Who likes marathons here?'
);
INSERT INTO posts_by_user (user_id, post_id, room_id, text) VALUES (
11111111-1111-1111-1111-111111111111,
aaaaaaaa-5cff-11ec-be16-1fedb0dfd057,
'#hiking', '... and Mt. Gumbo was easy!!!'
);
INSERT INTO posts_by_user (user_id, post_id, room_id, text) VALUES (
55555555-5555-5555-5555-555555555555,
bbbbbbbb-5cff-11ec-be16-1fedb0dfd057,
'#hiking', 'For us humans Gumbo is a tough one...!'
);
INSERT INTO posts_by_user (user_id, post_id, room_id, text) VALUES (
99999999-9999-9999-9999-999999999999,
cccccccc-5cff-11ec-be16-1fedb0dfd057,
'#running', 'I just love marathons.'
);
INSERT INTO posts_by_user (user_id, post_id, room_id, text) VALUES (
11111111-1111-1111-1111-111111111111,
eeeeeeee-5cff-11ec-be16-1fedb0dfd057,
'#running', 'Same here!'
);
INSERT INTO posts_by_user (user_id, post_id, room_id, text) VALUES (
55555555-5555-5555-5555-555555555555,
ffffffff-5cff-11ec-be16-1fedb0dfd057,
'#hiking', 'I have to buy new boots.'
);
Ok, we have a lovely bunch of posts in our chat application.
But wait: data is denormalized and the very same posts have to be inserted
in table posts_by_room as well! Let's do it with the following command
(please note that the INSERT
statements are exactly the same as above,
with only the table name changed):
π Commands to execute
// Insert some data in the "posts_by_room" table
INSERT INTO posts_by_room (user_id, post_id, room_id, text) VALUES (
11111111-1111-1111-1111-111111111111,
22222222-5cff-11ec-be16-1fedb0dfd057,
'#hiking', 'I climbed Mt. Gumbo yesterday ...'
);
INSERT INTO posts_by_room (user_id, post_id, room_id, text) VALUES (
11111111-1111-1111-1111-111111111111,
77777777-5cff-11ec-be16-1fedb0dfd057,
'#running', 'Who likes marathons here?'
);
INSERT INTO posts_by_room (user_id, post_id, room_id, text) VALUES (
11111111-1111-1111-1111-111111111111,
aaaaaaaa-5cff-11ec-be16-1fedb0dfd057,
'#hiking', '... and Mt. Gumbo was easy!!!'
);
INSERT INTO posts_by_room (user_id, post_id, room_id, text) VALUES (
55555555-5555-5555-5555-555555555555,
bbbbbbbb-5cff-11ec-be16-1fedb0dfd057,
'#hiking', 'For us humans Gumbo is a tough one...!'
);
INSERT INTO posts_by_room (user_id, post_id, room_id, text) VALUES (
99999999-9999-9999-9999-999999999999,
cccccccc-5cff-11ec-be16-1fedb0dfd057,
'#running', 'I just love marathons.'
);
INSERT INTO posts_by_room (user_id, post_id, room_id, text) VALUES (
11111111-1111-1111-1111-111111111111,
eeeeeeee-5cff-11ec-be16-1fedb0dfd057,
'#running', 'Same here!'
);
INSERT INTO posts_by_room (user_id, post_id, room_id, text) VALUES (
55555555-5555-5555-5555-555555555555,
ffffffff-5cff-11ec-be16-1fedb0dfd057,
'#hiking', 'I have to buy new boots.'
);
β Step 3c. C(R)UD = read = read data
Now that we've inserted a set of rows (two sets, to be precise), let's take a look at how to read the data back out. This is done with a SELECT statement. In its simplest form we could just execute a statement like the following **cough **cough:
// Read all rows from "posts_by_user" table (careful with this ...)
SELECT * FROM posts_by_user;
You may have noticed my coughing fit a moment ago. Even though you can execute a SELECT statement with no partition key defined, this is NOT something you should do when using Apache Cassandra. We are doing it here for illustration purposes only and because our whole dataset is just a handful of values. Given the data we inserted earlier, a more proper statement would be something like (while we are at it, we also explicitly specify which columns we want back):
// Read (some columns of) rows in a certain partition of "posts_by_user" table
SELECT post_id, room_id, text FROM posts_by_user
WHERE user_id = 11111111-1111-1111-1111-111111111111;
The key is to ensure we are always selecting by some partition key at a minimum, so to avoid the dreaded full-cluster scans which yield performances that are generally unacceptable in production.
Ok, with that out of the way we can READ the data from the other table as well - remember we INSERTed on both tables?
π Commands to execute
// Read the whole "posts_by_room" table
// (warning: not suitable for large tables in production)
SELECT * FROM posts_by_room;
// Read (some columns of) posts from a certain room (= a certain partition)
SELECT user_id, text FROM posts_by_room WHERE room_id = '#hiking';
(again, in the second SELECT we specify some columns - it is something we may want to do in most cases).
π Expected output
Notice how the two tables contain the same set of posts, but group them differently:
table posts_by_user
is partitioned by user, while table posts_by_room
is partitioned by room - and the corresponding outputs
reflect this fact.
This is very much related to the fact that these two tables, in the data modeling process, were designed
to answer two different questions, "what are the posts by user X?" and "what are the posts in room Y?" respectively.
Moreover, within any partition in both tables, right as we required when creating the table,
posts are kept (and displayed) sorted by decreasing post_id
(which, due to the nature of TIMEUUID
s,
implies a time-ordering as well).
Once you execute the above SELECT statements you should see something like the expected output above. We have now READ the data we INSERTED earlier. Awesome job!
BTW, just a little extra for those who are interested. Since we used a TIMEUUID type for our post_id field we can use the dateOf() function to determine the timestamp from the value. Check it out.
// Read all data from the posts_by_room table,
// convert post_id into a timestamp, and label the column "post_date"
SELECT user_id, dateOf(post_id) AS post_date, text FROM posts_by_room
WHERE room_id = '#hiking';
β Step 3d. CR(U)D = update = update data
At this point we've CREATED and READ some data, but what happens when you want to change some existing data to some new value? That's where UPDATE comes into play. The use case is as follows: in our chat app, users are allowed to edit their previous posts.
Let's take one of the records we created earlier and modify it. Recall that we INSERTED the following record in the posts_by_user table.
// ** Just for reference: **
// INSERT INTO posts_by_user (user_id, post_id, room_id, text) VALUES (
// 11111111-1111-1111-1111-111111111111,
// aaaaaaaa-5cff-11ec-be16-1fedb0dfd057,
// '#hiking', '... and Mt. Gumbo was easy!!!'
// );
Let's also take a look at how the posts_by_user table was created. In order to UPDATE an existing record, indeed, we need to know the primary key we defined when we CREATEd the table.
// ** Just for reference: **
// CREATE TABLE IF NOT EXISTS posts_by_user (
// user_id UUID,
// post_id TIMEUUID,
// room_id TEXT,
// text TEXT,
// PRIMARY KEY ((user_id), post_id)
// ) WITH CLUSTERING ORDER BY (post_id DESC);
Let's say that user "111..." has noticed the remark by "555..." and, perhaps a bit ashamed by their own boasting, wants to correct their assessment on the hike difficulty!
Looking at PRIMARY KEY ((user_id), post_id)
, we know that both user_id and post_id are used to define uniqueness of the row.
We'll need both to update our record (plus, of course, some of the data columns, otherwise we are not changing anything in that row!).
You may remember that we used hardcoded values for post_id when we created these records (a real application would generate them live, one way or the other). Imagine the UX for editing an existing post: when the user clicks the "edit" button, both user_id and post_id are known and can be provided to the backend, where they ultimately become part of an UPDATE statement.
So we can run the following UPDATE statement and help user "111..." fix their post on table posts_by_user (we also subsequently read back the data as a check):
π Commands to execute
UPDATE posts_by_user
SET text = '... and Mt. Gumbo was NOT SO easy!!!'
WHERE user_id = 11111111-1111-1111-1111-111111111111
AND post_id = aaaaaaaa-5cff-11ec-be16-1fedb0dfd057;
SELECT post_id, room_id, text FROM posts_by_user
WHERE user_id = 11111111-1111-1111-1111-111111111111;
π Expected output
But wait: data, again, is denormalized! This means that we have to make sure
such an edit is performed on table posts_by_room as well.
Since the primary key of that table is given as PRIMARY KEY ((room_id), post_id)
,
these are the fields to provide, along with text
itself, to the UPDATE statement.
And we could run an UPDATE. But, lo and behold, in Cassandra UPDATEs and INSERTs are (almost) the same, as a consequence of its architecture and the way storage and write logic are structured. We can then update the row with an INSERT statement like the following (note that we provide: primary key + any field that we want to modify; and leave out the other, unchanged fields):
π Commands to execute
INSERT INTO posts_by_room (room_id, post_id, text) VALUES (
'#hiking',
aaaaaaaa-5cff-11ec-be16-1fedb0dfd057,
'... and Mt. Gumbo was NOT SO easy!!!'
);
SELECT post_id, user_id, text FROM posts_by_room WHERE room_id = '#hiking';
That's it, we successfully edited a post (on both tables). All that's left now is to DELETE some data.
β Step 3e. CRU(D) = delete = remove data
The final operation from our CRUD acronym is DELETE. This is the operation we use when we want to remove data from the database. In Apache Cassandra you can DELETE from the cell level all the way up to the partition (meaning I could remove a single column in a single row or I could remove a whole partition) using the same DELETE command.
Generally speaking, it's best to perform as few delete operations as possible on the largest amount of data. Think of it this way, if you want to delete ALL data in a table, don't delete each individual cell, just TRUNCATE the table. If you need to delete all the rows in a partition, don't delete each row, DELETE the partition, and so on.
User "555..." notices the post by "111..." being edited and wants to remove their snarky remark. Let's help them!
When deleting a row on a given table, we have to specify the values of the primary key for that table. And don't forget that, in our data model, a post appears as two separate rows in the two tables, so we have to perform two different DELETE operations!
π Commands to execute
SELECT post_id, room_id, text FROM posts_by_user
WHERE user_id = 55555555-5555-5555-5555-555555555555;
SELECT post_id, user_id, text FROM posts_by_room WHERE room_id = '#hiking';
DELETE FROM posts_by_user
WHERE user_id = 55555555-5555-5555-5555-555555555555
AND post_id = bbbbbbbb-5cff-11ec-be16-1fedb0dfd057;
DELETE FROM posts_by_room
WHERE room_id = '#hiking'
AND post_id = bbbbbbbb-5cff-11ec-be16-1fedb0dfd057;
SELECT post_id, room_id, text FROM posts_by_user
WHERE user_id = 55555555-5555-5555-5555-555555555555;
SELECT post_id, user_id, text FROM posts_by_room WHERE room_id = '#hiking';
(Notice in the above, for your convenience, we read the tables, then delete the rows, then read them again).
π Expected output
Notice the rows are now removed from both tables: it is as simple as that.
To submit the homework, please take a screenshot of the CQL Console showing the rows in tables
posts_by_user
and posts_by_room
before and after executing the DELETE statements.
We've just scratched the surface of what you can do using Astra DB, built on Apache Cassandra. Go take a look at DataStax for Developers to see what else is possible. There's plenty to dig into!
Congratulations: you made to the end of today's workshop.
Don't forget to submit your homework and be awarded a nice verified badge!
... and see you at our next workshop!
Sincerely yours, The DataStax Developers