Implementation of Atomic Transactions in dynamodb - amazon-web-services

I have a table in dynamodb, where I need to update multiple related items at once(I can't put all data in one item because of 400kb size limit).
How can I make sure that either multiple rows are updated successfully or none.
End goal is to read consistent data after update.

On November 27th, 2018, transactions for Dynamo DB were announced. From the linked article:
DynamoDB transactions provide developers atomicity, consistency, isolation, and durability (ACID) across one or more tables within a single AWS account and region. You can use transactions when building applications that require coordinated inserts, deletes, or updates to multiple items as part of a single logical business operation. DynamoDB is the only non-relational database that supports transactions across multiple partitions and tables.
The new APIs are:
TransactWriteItems, a batch operation that contains a write set, with one or more PutItem, UpdateItem, and DeleteItem operations. TransactWriteItems can optionally check for prerequisite conditions that must be satisfied before making updates. These conditions may involve the same or different items than those in the write set. If any condition is not met, the transaction is rejected.
TransactGetItems, a batch operation that contains a read set, with one or more GetItem operations. If a TransactGetItems request is issued on an item that is part of an active write transaction, the read transaction is canceled. To get the previously committed value, you can use a standard read.
The linked article also has a JavaScript example:
data = await dynamoDb.transactWriteItems({
TransactItems: [
{
Update: {
TableName: 'items',
Key: { id: { S: itemId } },
ConditionExpression: 'available = :true',
UpdateExpression: 'set available = :false, ' +
'ownedBy = :player',
ExpressionAttributeValues: {
':true': { BOOL: true },
':false': { BOOL: false },
':player': { S: playerId }
}
}
},
{
Update: {
TableName: 'players',
Key: { id: { S: playerId } },
ConditionExpression: 'coins >= :price',
UpdateExpression: 'set coins = coins - :price, ' +
'inventory = list_append(inventory, :items)',
ExpressionAttributeValues: {
':items': { L: [{ S: itemId }] },
':price': { N: itemPrice.toString() }
}
}
}
]
}).promise();

You can use an API like this one for Java, http://aws.amazon.com/blogs/aws/dynamodb-transaction-library/. The transaction library API will help you manage atomic transactions.
If you're using node.js, there are other solutions for that using an atomic counter or conditional writes. See answer here, How to support transactions in dynamoDB with javascript aws-sdk?.

Related

DynamoDB batchGet vs multiple getItem

Is there any difference in terms of cost or performance between using multiple getItem calls:
Promise.all([
client.getItem({ TableName, Key }).promise(),
client.getItem({ TableName, Key }).promise(),
client.getItem({ TableName, Key }).promise(),
]
to one batchGet call:
const params = {
RequestItems: {
'TABLE_NAME': {
Keys: [
{'KEY_NAME': {N: 'KEY_VALUE_1'}},
{'KEY_NAME': {N: 'KEY_VALUE_2'}},
{'KEY_NAME': {N: 'KEY_VALUE_3'}}
]
}
}
};
db.batchGetItem(params).promise()
In terms of cost: No - both Operations consume the same number of Read Capacity Units.
In terms of performance: Yes - using multiple GetItem requests sends separate network requests for each of them and for BatchGetItem there is only one request, which should be quite a bit faster.
There is no real downside to BatchGetItem except for a slight increase in complexity when some items aren't found.

Not able to solve throttlingException in DynamoDB

I have a lambda function which does a transaction in DynamoDB similar to this.
try {
const reservationId = genId();
await transactionFn();
return {
statusCode: 200,
body: JSON.stringify({id: reservationId})
};
async function transactionFn() {
try {
await docClient.transactWrite({
TransactItems: [
{
Put: {
TableName: ReservationTable,
Item: {
reservationId,
userId,
retryCount: Number(retryCount),
}
}
},
{
Update: {
TableName: EventDetailsTable,
Key: {eventId},
ConditionExpression: 'available >= :minValue',
UpdateExpression: `set available = available - :val, attendees= attendees + :val, lastUpdatedDate = :updatedAt`,
ExpressionAttributeValues: {
":val": 1,
":updatedAt": currentTime,
":minValue": 1
}
}
}
]
}).promise();
return true
} catch (e) {
const transactionConflictError = e.message.search("TransactionConflict") !== -1;
// const throttlingException = e.code === 'ThrottlingException';
console.log("transactionFn:transactionConflictError:", transactionConflictError);
if (transactionConflictError) {
retryCount += 1;
await transactionFn();
return;
}
// if(throttlingException){
//
// }
console.log("transactionFn:e.code:", JSON.stringify(e));
throw e
}
}
It just updating 2 tables on api call. If it encounter a transaction conflict error, it simply retry the transaction by recursively calling the function.
The eventDetails table is getting too much db updates. ( checked it with aws Contributor Insights) so, made provisioned unit to a higher value than earlier.
For reservationTable Provisioned capacity is on Demand.
When I do load test over this api with 400 (or more) users using JMeter (master slave configuration) I am getting Throttled error for some api calls and some api took more than 20 sec to respond.
When I checked X-Ray for this api found that, DynamoDB is taking too much time for this transasction for the slower api calls.
Even with much fixed provisioning ( I tried on demand scaling too ) , I am getting throttled exception for api calls.
ProvisionedThroughputExceededException: The level of configured provisioned throughput for the table was exceeded.
Consider increasing your provisioning level with the UpdateTable API.
UPDATE
And one more thing. When I do the load testing, I am always uses the same eventId. It means, I am always updating the same row for all the api requests. I have found this article, which says that, a single partition can only have upto 1000 WCU. Since I am always updating the same row in the eventDetails table during load testing, is that causing this issue ?
I had this exact error and it helped me to change the On Demand to Provisioned under Read/write capacity mode. Try to change that, if that doesn't help, we'll go from there.
From the link you cite in your update, also described in an AWS help article here, it sounds like the issue is that all of your load testers are writing to the same entry in the table, which is going to be in the same partition, subject to the hard limit of 1,000 WCU.
Have you tried repeating this experiment with the load testers writing to different partitions?

Map different Sort Key responses to Appsync Schema values

So here is my schema:
type Model {
PartitionKey: ID!
Name: String
Version: Int
FBX: String
# ms since epoch
CreatedAt: AWSTimestamp
Description: String
Tags: [String]
}
type Query {
getAllModels(count: Int, nextToken: String): PaginatedModels!
}
type PaginatedModels {
models: [Model!]!
nextToken: String
}
I would like to call 'getAllModels' and have all of it's data, and all of it's tags be filled in.
But here is the thing. Tags are stored via sort keys. Like so
PartionKey | SortKey
Model-0 | Model-0
Model-0 | Tag-Tree
Model-0 | Tag-Building
Is it possible to transform the 'Tag' sort keys into the Tags: [String] array in the schema via a DynamoDB resolver? Or must I do something extra fancy through a lambda? Or is there a smarter way to do this?
To clarify, are you storing objects like this in DynamoDB:
{ PartitionKey (HASH), Tag (SortKey), Name, Version, FBX, CreatedAt, Description }
and using a DynamoDB Query operation to fetch all rows for a given HashKey.
Query #PartitionKey = :PartitionKey
and getting back a list of objects some of which have a different "Tag" value and one of which is "Model-0" (aka the same value as the partition key) and I assume that record contains all other values for the record. E.G.
[
{ PartitionKey, Tag: 'ValueOfPartitionKey', Name, Version, FBX, CreatedAt, ... },
{ PartitionKey, Tag: 'Tag-Tree' },
{ PartitionKey: Tag: 'Tag-Building' }
]
You can definitely write resolver logic without too much hassle that reduces the list of model objects into a single object with a list of "Tags". Let's start with a single item and see how to implement a getModel(id: ID!): Model query:
First define the response mapping template that will get all rows for a partition key:
{
"version" : "2017-02-28",
"operation" : "Query",
"query" : {
"expression": "#PartitionKey = :id",
"expressionValues" : {
":id" : {
"S" : "${ctx.args.id}"
}
},
"expressionNames": {
"#PartitionKey": "PartitionKey" # whatever the table hash key is
}
},
# The limit will have to be sufficiently large to get all rows for a key
"limit": $util.defaultIfNull(${ctx.args.limit}, 100)
}
Then to return a single model object that reduces "Tag" to "Tags" you can use this response mapping template:
#set($tags = [])
#set($result = {})
#foreach( $item in $ctx.result.items )
#if($item.PartitionKey == $item.Tag)
#set($result = $item)
#else
$util.qr($tags.add($item.Tag))
#end
#end
$util.qr($result.put("Tags", $tags))
$util.toJson($result)
This will return a response like this:
{
"PartitionKey": "...",
"Name": "...",
"Tags": ["Tag-Tree", "Tag-Building"],
}
Fundamentally I see no problem with this but its effectiveness depends upon your query patterns. Extending this to the getAll use is doable but will require a few changes and most likely a really inefficient Scan operation due to the fact that the table will be sparse of actual information since many records are effectively just tags. You can alleviate this with GSIs pretty easily but more GSIs means more $.
As an alternative approach, you can store your Tags in a different "Tags" table. This way you only store model information in the Model table and tag information in the Tag table and leverage GraphQL to perform the join for you. In this approach have Query.getAllModels perform a "Scan" (or Query) on the Model table and then have a Model.Tags resolver that performs a Query against the Tag table (HK: ModelPartitionKey, SK: Tag). You could then get all tags for a model and later create a GSI to get all models for a tag. You do need to consider that now the nested Model.Tag query will get called once per model but Query operations are fast and I've seen this work well in practice.
Hope this helps :)

Access to AWS Kinesis tables in DynamoDB using AWS Cli

I am new to AWS Kinesis. Trying to learn how to build distributed application using AWS DynamoDB. Could someone tell me how to access the tables in DynamoDB used by my streams in AWS Kinesis using the AWS Cli? Can I query the tables?
Surely. Kinesis-stream uses Amazon Dynamodb table for consumer offset management.
your table will have the same name as your consumer name (defined via KCL). you can see all consumer_offset_tables via console (https://console.aws.amazon.com/dynamodb/home)
consumer could be running in multiple instances, but only one instance (THE leaseOwner)can consume from one partition (or Shard what Kinesis calls it). If this consumer fails another instance of the same consumer will take over and continue processing from the checkpoint.
checkpoint is the last processed event
the shard that the consumer instance is processing is called leaseKey which is unique for a table.
The data structure of the key value based document in Dynamodb would be as below,
where
"S" - Char array
"N" - Number
{
"leaseOwner": {
"S": "SmartConsumerStream_Consumer-192.168.1.83"
},
"checkpoint": {
"S": "49570630332110756564477900867375857710984404992079691778"
},
"checkpointSubSequenceNumber": {
"N": "0"
},
"leaseCounter": {
"N": "16"
},
"leaseKey": {
"S": "shardId-000000000000"
},
"ownerSwitchesSinceCheckpoint": {
"N": "0"
}
}
You can use Dynamodb API to get the current offset for given partitionKey or leaseKey. You can only query by leaseKey because thats the indexed key in table. Its created by Kinesis-stream itself.
You can use the stream-driver I'm writing at here, which gives you interface to get the consumer offset very very easily.
Here are kinesis-stream tests which might be helpful too.
In summary, get the consumer offset using JAVA api
-- put aws credentials in ~/.aws/credentials
public Map<String, String> getConsumerPosition() {
DynamoDB dynamoDB = new DynamoDB(getOffsetConnection()); //
Table consumerOffsetTable = dynamoDB.getTable("your_consumer_id");
Map<String, Object> leaseOwner = consumerOffsetTable.getItem("leaseKey", "shardId-000000000000").asMap();
return new HashMap<String, String>(){{
put(leaseOwner.get("leaseKey").toString(), leaseOwner.get("checkpoint").toString());
}};
}
public AmazonDynamoDB getOffsetConnection() {
AmazonDynamoDBClient dynamoDB = new AmazonDynamoDBClient(getAuthProfileCredentials(), getHttpConfiguration());
return dynamoDB;
}
private ProfileCredentialsProvider getAuthProfileCredentials() {
return new ProfileCredentialsProvider("your-aws-auth-profile_in_~/.aws/credentials");
}
private ClientConfiguration getHttpConfiguration() {
ClientConfiguration clientConfiguration = new ClientConfiguration();
return clientConfiguration;
}
Hope it helps, let me know if I can help some other ways.

Copying one table to another in DynamoDB

What's the best way to identically copy one table over to a new one in DynamoDB?
(I'm not worried about atomicity).
Create a backup(backups option) and restore the table with a new table name. That would get all the data into the new table.
Note: Takes considerable amount of time depending on the table size
I just used the python script, dynamodb-copy-table, making sure my credentials were in some environment variables (AWS_ACCESS_KEY_ID and AWS_SECRET_ACCESS_KEY), and it worked flawlessly. It even created the destination table for me.
python dynamodb-copy-table.py src_table dst_table
The default region is us-west-2, change it with the AWS_DEFAULT_REGION env variable.
AWS Pipeline provides a template which can be used for this purpose: "CrossRegion DynamoDB Copy"
See: http://docs.aws.amazon.com/datapipeline/latest/DeveloperGuide/dp-crossregion-ddb-create.html
The result is a simple pipeline that looks like:
Although it's called CrossRegion you can easily use it for the same region as long the destination table name is different (Remember that table names are unique per account and region)
You can use Scan to read the data and save it to the new table.
On the AWS forums a guy from the AWS team posted another approach using EMR: How Do I Duplicate a Table?
Here's one solution to copy all items from one table to another, just using shell scripting, the AWS CLI and jq. Will work OK for smallish tables.
# exit on error
set -eo pipefail
# tables
TABLE_FROM=<table>
TABLE_TO=<table>
# read
aws dynamodb scan \
--table-name "$TABLE_FROM" \
--output json \
| jq "{ \"$TABLE_TO\": [ .Items[] | { PutRequest: { Item: . } } ] }" \
> "$TABLE_TO-payload.json"
# write
aws dynamodb batch-write-item --request-items file://"$TABLE_TO-payload.json"
# clean up
rm "$TABLE_TO-payload.json"
If you both tables to be identical, you'd want to delete all items in TABLE_TO first.
DynamoDB now supports importing from S3.
https://aws.amazon.com/blogs/database/amazon-dynamodb-can-now-import-amazon-s3-data-into-a-new-table/
So, probably in almost all use cases, the easiest and cheapest way to replicate a table is
Use "Export to S3" feature to dump entire table into S3. Since this uses backup to generate the dump, table's throughput is not affected, and is very fast as well. You need to have backups (PITR) enabled. See https://aws.amazon.com/blogs/aws/new-export-amazon-dynamodb-table-data-to-data-lake-amazon-s3/
Use "Import from S3" to import the dump created in step 1. This automatically requires you to create a new table.
Use this node js module : copy-dynamodb-table
This is a little script I made to copy the contents of one table to another.
It's based on the AWS-SDK v3. Not sure how well it would scale to big tables but as a quick and dirty solution it does the job.
It gets your AWS credentials from a profile in ~/.aws/credentials change default to the name of the profile you want to use.
Other than that it takes two args one for the source table and one for destination
const { fromIni } = require("#aws-sdk/credential-providers");
const { DynamoDBClient, ScanCommand, PutItemCommand } = require("#aws-sdk/client-dynamodb");
const ddbClient = new DynamoDBClient({
credentials: fromIni({profile: "default"}),
region: "eu-west-1",
});
const args = process.argv.slice(2);
console.log(args)
async function main() {
const { Items } = await ddbClient.send(
new ScanCommand({
TableName: args[0],
})
);
console.log("Successfully scanned table")
console.log("Copying", Items.length, "Items")
const putPromises = [];
Items.forEach((item) => {
putPromises.push(
ddbClient.send(
new PutItemCommand({
TableName: args[1],
Item: item,
})
)
);
});
await Promise.all(putPromises);
console.log("Successfully copied table")
}
main();
Usage
node copy-table.js <source_table_name> <destination_table_name>
Python + boto3 🚀
The script is idempotent as far as you maintain the same Keys.
import boto3
def migrate(source, target):
dynamo_client = boto3.client('dynamodb', region_name='us-east-1')
dynamo_target_client = boto3.client('dynamodb', region_name='us-west-2')
dynamo_paginator = dynamo_client.get_paginator('scan')
dynamo_response = dynamo_paginator.paginate(
TableName=source,
Select='ALL_ATTRIBUTES',
ReturnConsumedCapacity='NONE',
ConsistentRead=True
)
for page in dynamo_response:
for item in page['Items']:
dynamo_target_client.put_item(
TableName=target,
Item=item
)
if __name__ == '__main__':
migrate('awesome-v1', 'awesome-v2')
On November 29th, 2017 Global Tables was introduced. This may be useful depending on your use case, which may not be the same as the original question. Here are a few snippets from the blog post:
Global Tables – You can now create tables that are automatically replicated across two or more AWS Regions, with full support for multi-master writes, with a couple of clicks. This gives you the ability to build fast, massively scaled applications for a global user base without having to manage the replication process.
...
You do not need to make any changes to your existing code. You simply send write requests and eventually consistent read requests to a DynamoDB endpoint in any of the designated Regions (writes that are associated with strongly consistent reads should share a common endpoint). Behind the scenes, DynamoDB implements multi-master writes and ensures that the last write to a particular item prevails. When you use Global Tables, each item will include a timestamp attribute representing the time of the most recent write. Updates are propagated to other Regions asynchronously via DynamoDB Streams and are typically complete within one second (you can track this using the new ReplicationLatency and PendingReplicationCount metrics).
Another option is to download the table as a .csv file and upload it with the following snippet of code.
This also eliminates the need for providing your AWS credentials to a packages such as the one #ezzat suggests.
Create a new folder and add the following two files and your exported table
Edit uploadToDynamoDB.js and add the filename of the exported table and your table name
Run npm install in the folder
Run node uploadToDynamodb.js
File: Package.json
{
"name": "uploadtodynamodb",
"version": "1.0.0",
"description": "",
"main": "uploadToDynamoDB.js",
"author": "",
"license": "ISC",
"dependencies": {
"async": "^3.1.1",
"aws-sdk": "^2.624.0",
"csv-parse": "^4.8.5",
"fs": "0.0.1-security",
"lodash": "^4.17.15",
"uuid": "^3.4.0"
}
}
File: uploadToDynamoDB.js
var fs = require('fs');
var parse = require('csv-parse');
var async = require('async');
var _ = require('lodash')
var AWS = require('aws-sdk');
// If your table is in another region, make sure to update this
AWS.config.update({ region: "eu-central-1" });
var ddb = new AWS.DynamoDB({ apiVersion: '2012-08-10' });
var csv_filename = "./TABLE_CSV_EXPORT_FILENAME.csv";
var tableName = "TABLENAME"
function prepareData(data_chunk) {
const items = data_chunk.map(obj => {
const keys = Object.keys(obj)
let attr = Object.values(obj)
attr = attr.map(a => {
let newAttr;
// Can we make this an integer
if (isNaN(Number(a))) {
newAttr = { "S": a }
} else {
newAttr = { "N": a }
}
return newAttr
})
let item = _.zipObject(keys, attr)
return {
PutRequest: {
Item: item
}
}
})
var params = {
RequestItems: {
[tableName]: items
}
};
return params
}
rs = fs.createReadStream(csv_filename);
parser = parse({
columns : true,
delimiter : ','
}, function(err, data) {
var split_arrays = [], size = 25;
while (data.length > 0) {
split_arrays.push(data.splice(0, size));
}
data_imported = false;
chunk_no = 1;
async.each(split_arrays, function(item_data, callback) {
const params = prepareData(item_data)
ddb.batchWriteItem(
params,
function (err, data) {
if (err) {
console.log("Error", err);
} else {
console.log("Success", data);
}
});
}, function() {
// run after loops
console.log('all data imported....');
});
});
rs.pipe(parser);
It's been a very long time since the question was posted and AWS has been continuously improvising features. At the time of writing this answer, we have the option to export the Table to S3 bucket then use the import feature to import this data from S3 into a new table which automatically will re-create a new table with the data. Plese refer this blog for more idea on export & import
Best part is that you get to change the name, PK or SK.
Note: You have to enable PITR (might incur additional costs). Always best to refer documents.
Here is another simple python util script for this: ddb_table_copy.py. I use it often.
usage: ddb_table_copy.py [-h] [--dest-table DEST_TABLE] [--dest-file DEST_FILE] source_table
Copy all DynamoDB items from SOURCE_TABLE to either DEST_TABLE or DEST_FILE. Useful for migrating data during a stack teardown/re-creation.
positional arguments:
source_table Name of source table in DynamoDB.
optional arguments:
-h, --help show this help message and exit
--dest-table DEST_TABLE
Name of destination table in DynamoDB.
--dest-file DEST_FILE
2) a valid file path string to save the items to, e.g. 'items.json'.