ChaNa

Avro Data Store based on Akka (persistence in progressing)

This project is named from Chinese word "刹那", which is a transliteration of the word "Kasna" from Sanskrit, means "instant; split second".

Core Design

• Each record is an actor (non-blocking)
• Akka sharding cluster (easy to scale-out)
• Locate field/value via XPath (since version 0.2.0)
• Scripting triggered by field updating events (JDK 8 JavaScript engine

  -Nashorn)

• JPQL query on cluster

Run chana

$ sbt run

Or

$ sbt clean compile dist
$ ls target/universal/
chana-0.2.0-SNAPSHOT.zip 

Then, copy chana-0.2.0-SNAPSHOT.zip to somewhere and unzip it

$ cd chana-0.2.0-SNAPSHOT/bin
$ ./chana

Access chana

Example 1: Simple Record

Schema: PersonInfo.avsc

{
  "type" : "record",
  "name" : "PersonInfo",
  "namespace" : "chana",
  "fields" : [ {
    "name" : "name",
    "type" : "string"
  }, {
    "name" : "age",
    "type" : "int"
  }, {
    "name" : "gender",
    "type" : {
      "type" : "enum",
      "name" : "GenderType",
      "symbols" : [ "Female", "Male", "Unknown" ]
    },
    "default" : "Unknown"
  }, {
    "name" : "emails",
    "type" : {
      "type" : "array",
      "items" : "string"
    }
  } ]
}

Try it:

$ cd src/test/resources/avsc

$ curl --data @PersonInfo.avsc 'http://127.0.0.1:8080/schema/put/personinfo'
OK

$ curl 'http://127.0.0.1:8080/personinfo/get/1'
{"name":"","age":0,"gender":"Unknown","emails":[]}

$ curl --data-binary @PersonInfo.update 'http://127.0.0.1:8080/personinfo/update/1'
OK

$ curl 'http://127.0.0.1:8080/personinfo/get/1'
{"name":"James Bond","age":60,"gender":"Unknown","emails":[]}

$ curl 'http://127.0.0.1:8080/personinfo/get/1/name'
"James Bond"

JPQL example

#### JPQL Simple test
$ cat jpql.put

echo 'SELECT COUNT(p.age), AVG(p.age) FROM PersonInfo p WHERE p.age >= 30 ORDER BY p.age' | \
curl -d @- 'http://127.0.0.1:8080/jpql/put/JPQL_NO_1'

$ ./jpql.put

#### watching jpql results
$ cat jpql.ask

while :
do
   sleep 1s
   curl 'http://127.0.0.1:8080/jpql/ask/JPQL_NO_1'
   printf '\n'
done

$ ./jpql.ask

#### update record with random id, repeat it to update more person's age to 80
$ cat jpql.update

RANDOM_ID=$RANDOM
JPQL="UPDATE PersonInfo p SET p.age = 80 WHERE p.id = '$RANDOM_ID'"
echo $JPQL

echo $JPQL | curl -d @- "http://127.0.0.1:8080/jpql"

$ ./jpql.update

Script example: (requires JDK8+)

A piece of JavaScript code that will be executed when field PersionInfo.name was updated: on_name.js:

function onNameUpdated() {
    var age = record.get("age");
    what_is(age);

    what_is(http_get);
    var http_get_result = http_get.apply("http://localhost:8080/ping", 5);
    java.lang.Thread.sleep(1000);
    what_is(http_get_result.value());

    what_is(http_post);
    var http_post_result = http_post.apply("http://localhost:8080/personinfo/put/2/age", "888", 5);
    java.lang.Thread.sleep(1000);
    what_is(http_post_result.value());

    for (i = 0; i < binlogs.length; i++) {
        var binlog = binglogs[i];
        what_is(binlog.type());
        what_is(binlog.xpath());
        what_is(binlog.value());
    }
}

function what_is(value) {
    print(id + ": " + value);
}

onNameUpdated();

Try it:

$ curl --data-binary @on_name.js \
 'http://127.0.0.1:8080/personinfo/script/put/name/SCRIPT_NO_1'
OK

$ curl --data '"John"' 'http://127.0.0.1:8080/personinfo/put/1/name'
OK

$ curl 'http://127.0.0.1:8080/personinfo/get/2/age'
888

Example 2: With Embedded Type

Schema: hatInventory.avsc

{
  "type" : "record",
  "name" : "hatInventory",
  "namespace" : "chana",
  "fields" : [ {
    "name" : "sku",
    "type" : "string",
    "default" : ""
  }, {
    "name" : "description",
    "type" : {
      "type" : "record",
      "name" : "hatInfo",
      "fields" : [ {
        "name" : "style",
        "type" : "string",
        "default" : ""
      }, {
        "name" : "size",
        "type" : "string",
        "default" : ""
      }, {
        "name" : "color",
        "type" : "string",
        "default" : ""
      }, {
        "name" : "material",
        "type" : "string",
        "default" : ""
      } ]
    },
    "default" : { }
  } ]
}

Try it:

$ cd src/test/resources/avsc

$ curl --data @hatInventory.avsc 'http://127.0.0.1:8080/schema/put/hatinv'
OK

$ curl 'http://127.0.0.1:8080/hatinv/get/1'
{"sku":"","description":{"style":"","size":"","color":"","material":""}}

$ curl --data '{"style":"classic","size":"Large","color":"Red"}' \
 'http://127.0.0.1:8080/hatinv/put/1/description'
OK

$ curl 'http://127.0.0.1:8080/hatinv/get/1'
{"sku":"","description":{"style":"classic","size":"Large","color":"Red","material":""}}

$ curl 'http://127.0.0.1:8080/hatinv/get/1/description'
{"style":"classic","size":"Large","color":"Red","material":""}

$ ab -c100 -n100000 -k 'http://127.0.0.1:8080/hatinv/get/1?benchmark_only=1024'

Simple benchmark for REST-JSON API (too simple too naive)

Environment:

HOST: Dell Inc. PowerEdge R420/0VD50G
CPU: 2 x Intel(R) Xeon(R) CPU E5-2420 v2 @ 2.20GHz (12 #core, 24 #HT)
OS: CentOS Linux release 7.0.1406 (Core)

Simple GET/PUT REST-JSON Result:

Simple GET: 169,437 [req#/sec] (mean)
Simple PUT: 102,961 [req#/sec] (mean)

Details:

• Benchmark
• Benchmark through multiple-core

To run:

sbt run
cd src/test/resources/avsc
./bench-get.sh
./bench-put.sh

Preface

chana stores Avro record, with two groups of APIs:

• Primitive API (Scala/Java)
• RESTful API

Primitive API (Scala / Java)

use XPath expression to locate avro field. See XPath

1. Schema

case class PutSchema(entityName: String, schema: String, entityFullName: Option[String], idleTimeout: Duration)
case class RemoveSchema(entityName: String)

2. Basic operations

case class GetRecord(id: String)
case class GetRecordAvro(id: String)
case class GetRecordJson(id: String)
case class PutRecord(id: String, record: Record)
case class PutRecordJson(id: String, record: String)
case class GetField(id: String, field: String)
case class GetFieldAvro(id: String, field: String)
case class GetFieldJson(id: String, field: String)
case class PutField(id: String, field: String, value: Any)
case class PutFieldJson(id: String, field: String, value: String)

case class Select(id: String, path: String)
case class SelectAvro(id: String, path: String)
case class SelectJson(id: String, path: String)
case class Update(id: String, path: String, value: Any)
case class UpdateJson(id: String, path: String, value: String)

3. Operations applicable on Array / Map

case class Insert(id: String, path: String, value: Any)
case class InsertJson(id: String, path: String, value: String)
case class InsertAll(id: String, path: String, values: List[_])
case class InsertAllJson(id: String, path: String, values: String)
case class Delete(id: String, path: String)
case class Clear(id: String, path: String)

4. Script

case class PutScript(entity: String, field: String, id: String, script: String)
case class RemoveScript(entity: String, field: String, id: String)

REST API

Put schema

POST /schema/put/$entityName?fullname=entity_full_name&timeout=1000

Host: status.wandoujia.com  
Content-Type: application/octet-stream 
Content-Length: NNN

BODY:
<SCHEMA_STRING>

parameters:

• fullname: for schema that contains multiple referenced complex types in union, you should provide the full name of main entry. Optional
• timeout: idle timeout in milliseconds. Optional

Del schema

GET /schema/del/$entityName/ 

Host: status.wandoujia.com  

Get record

GET /$entity/get/$id/ 

Host: status.wandoujia.com  

Get record field

GET /$entity/get/$id/$field

Host: status.wandoujia.com  

Put record

POST /$entity/put/$id/ 

Host: status.wandoujia.com  
Content-Type: application/octet-stream 
Content-Length: NNN

BODY:
<JSON_STRING>

Put record field

POST /$entity/put/$id/$field 

Host: status.wandoujia.com  
Content-Type: application/octet-stream 
Content-Length: NNN

BODY:
<JSON_STRING>

Select

POST /$entity/select/$id/ 

Host: status.wandoujia.com  
Content-Type: application/octet-stream 
Content-Length: NNN

BODY:
$xpath

Update

POST /$entity/update/$id/

Host: status.wandoujia.com 
Content-Type: application/octet-stream 
Content-Length: NNN

BODY:
$xpath
<JSON_STRING>

Example (update array field -> record’s number field):

POST /account/update/12345/
BODY: 
/chargeRecords[1].time
1234

Example (update map field -> record’s number field):

POST /account/update/12345/
BODY:
/devApps/@a/numBlackApps
1234

Insert (applicable for Array / Map only)

POST /$entity/insert/$id/

Host: status.wandoujia.com 
Content-Type: application/octet-stream 
Content-Length: NNN

BODY:
$xpath
<JSON_STRING>

Example (insert to array field):

POST /account/insert/12345/
BODY: 
/chargeRecords
{"time": 4, "amount": -4.0}

Example (insert to map field):

POST /account/insert/12345/
BODY: 
/devApps
{"h" : {"numBlackApps": 10}}

InsertAll (applicable for Array / Map only)

POST /$entity/insertall/$id/

Host: status.wandoujia.com 
Content-Type: application/octet-stream 
Content-Length: NNN

BODY:
$xpath
<JSON_STRING>

Example (insert to array field):

POST /account/insertall/12345/
BODY: 
/chargeRecords
[{"time": -1, "amount": -5.0}, {"time": -2, "amount": -6.0}]

Example (insert to map field):

POST /account/insertall/12345/
BODY: 
/devApps
{"g" : {}, "h" : {"numBlackApps": 10}}

Delete (applicable for Array / Map only)

POST /$entity/delete/$id/

Host: status.wandoujia.com 
Content-Type: application/octet-stream 
Content-Length: NNN

BODY:
$xpath

Clear (applicable for Array / Map only)

POST /$entity/clear/$id/

Host: status.wandoujia.com 
Content-Type: application/octet-stream 
Content-Length: NNN

BODY:
$xpath

Put Script (apply on all instances of this entity)

POST /$entity/script/put/$field/$scriptid/

Host: status.wandoujia.com 
Content-Type: application/octet-stream 
Content-Length: NNN

BODY:
<JavaScript>

Del Script (apply on all instances of this entity)

GET /$entity/script/del/$field/$scriptid/

Host: status.wandoujia.com 

Note:

• Replace $entity with the object/table/entity name
• Replace $id with object id
• Replace $xpath with actual xpath expression
• Put the $xpath and JSON format value(s) for update / insert / insertall in POST body, separate $xpath and JSON value(s) with \n, and make sure it’s encoded as binary, set Content-Type: application/octet-stream

Scripting supporting

The bindings that could be accessed in script:

def prepareBindings(onUpdated: OnUpdated) = {
  val bindings = new SimpleBindings
  bindings.put("http_get", http_get)
  bindings.put("http_post", http_post)
  bindings.put("id", onUpdated.id)
  bindings.put("record", record)
  bindings.put("binlogs", binlogs)
  bindings
}

Where,

• http_get: a function could be invoked via http_get.apply(url: CharSequence, timeoutInSeconds: Int), returns scala.concurrent.Future[Any]
• http_post: a function could be invoked via http_post.apply(url: CharSequence, body: CharSequence, timeoutInSeconds: Int) returns scala.concurrent.Future[Any]
• id: the id of this entity
• record: the entity record after updated
• binlogs: array of Binlog(s) during this updating action
• binlogs[i].type(): -1 - Clear, 0 - Delete, 1 - Change, 2 - Insert
• binlogs[i].xpath(): XPath Location of changed value
• binlogs[i].value(): value - Change/Insert, keys - Delete, void - Clear
• The JavaScript code should do what ever operation via function only. You can define local variables in function, and transfer these local vars between functions to share them instead of defining global vars.

Reference

• XPath
• Nashorn