Minimal

Efficiently process a sequence of integers to determine the minimal value

Design

This is not a production-grade solution, but demonstrates the use of streaming to process
a large data set without a corresponding - and deleterious - increase is memory footprint

Configuration

The maximum length of allowed frames (in this case, discerete integer values) while decoding, set in conf/application.conf:

input.frame.size=1000

API

Calculate minimum value

Given a data-set consisting of a set of comma-seperated Long integer values (positive and negative), determine and answer with the lowest value

URL : /api/minimal

Method : POST

Auth required : No

Permissions required : None

Data constraints

Provide comma-separated list of integer values in the domain of java.lang.Long

Data example

-994275329,
906862559,
-36626088,
743799515,
-201965995,
-21821772,
322782558,
-422943259,
-956974090

Success Response

Condition : If data set processed ok

Code : 200 SUCCESS

Content example

-999968087

Error Responses

Condition : If empty data-set provided

Code : 400 BAD REQUEST

Content example : no data provided

Run Service

Start the application locally:

% sbt run

Send data to the api:

% curl  http://localhost:9000/api/minimal -H "Content-type: multipart/form-data" -d @test/resources/bigfile.dat

This will respond with the minimum value from the data-set:

-999968087

Metrics

Retrieve JVM metrics

API to retrieve JVM metrics values from running service (@codahale)

URL : /metrics

Method : GET

Auth required : No

Permissions required : None

Data example

{
  "version" : "4.0.0",
  "gauges" : {
    "jvm.attribute.name" : {
      "value" : "73662@PDQ17526984"
    },
    "jvm.attribute.uptime" : {
      "value" : 150893
    },
    "jvm.attribute.vendor" : {
      "value" : "Oracle Corporation Java HotSpot(TM) 64-Bit Server VM 25.192-b12 (1.8)"
    }
  },
  "timers" : { }
}

Testing

The provided unit tests verify that integer values in the domain
-1000000000 thru 1000000000 are correctly processed, and invalid values
filtered out of the processing. A data-set of 9999 randomly-generated values
is used, see test/resources/bigfile.dat:

sort --numeric-sort bigfile.dat | head 
-999968087
-999800455
-999597000
-999008584
-998948663
-998653326
-998463590
-998305391
-998182735
-999968087
-998018946

Author

• Roy Phillips