#URL Tracker Example
This application manages customer data, user data and site-visit data. The purpose of this example application is to highlight different ways to employ Aerospike Large Data Types (LDTs) to manage collections of objects -- in this case, the site-visit data. This example shows two different ways to manage a collection of "site-visit" values that correspond to user data; one method uses a Large Ordered List, and the other method uses a Large Map.
In the Aerospike Database, customer data is kept in a set. There are many sets -- one for each customer. In each customer set, there are user records. Each user record describes a user. In each user record, besides the user data, there is a collection of site-visit information that pertains to a user. The site-visit data collection is managed by an Aerospike Large Data Type (LDT).
The overall data schema is as follows:
* Overview:
* There are many User Records per Customer Set
Each User Data Record contains:
- Key (user id)
- User Info
- Site Visit Data List: Each list item is a map, containing:
(URL, Referrer, Page Title, AgentID, IP Addr, Date, Expire, etc)
+--------------------------------------------------------------+
| CUSTOMER SET THREE |
+-------------------------------------------------------------+ |
| CUSTOMER SET TWO | |
+------------------------------------------------------------+ | |
| CUSTOMER SET ONE | | |
| Singleton [ Customer Data ] | | |
| User Records | | |
| +-------------------------------------------------+ | | |
| +-------------------------------------------------+| | | |
| +-------------------------------------------------+|| | | |
| | | | | ||| | | |
| | Key | User Info | Site Visit Data List | o o o ||+ | | |
| | | | | |+ | | |
| +--------------|-----------------------|----------+ | | |
| | V | | |
| V +---------------+ | | |
| +------------+ +---------------+| | | |
| |Info Data | +---------------+|| | | |
| |------------| |Site Visit Data||| | | |
| |* User Name | |---------------||| | | |
| |* Email | |* URL ||| | | |
| |* Cell phone| |* Referrer ||| | | |
| |* Address | |* Page Title ||| | | |
| |* Company | |* UserAgent-ID ||| | | |
| |* o o o | |* IP Address ||| | |----+
| |* etc | |* Date ||+ | |
| +------------+ |* Expire |+ |----+
| +---------------+ |
+------------------------------------------------------------+
In this example application, we show two different ways to manage the site-visit data. It can be managed by a Large List, where the ordering criteria is the expire value; or, it can be managed by a a Large Map, where the expire value is the lookup value.
There are five main operations:
- Add a new Customer Record
- Add a new user Record
- For an existing user, add a new Site Visit Object to the collection
- Query the Site Visit Object (with filters on object fields)
- Remove all entries that have an expire time < current time.
There are several additional (supportive) operations:
- Query all records in the customer set
- Remove a record (by key)
- Remove all records in a customer set
##Problem
The two main problems addressed by the example application are:
- How do we represent the Site-Visit Object list?
- How do we expire site values that are older than a given value?
##Solution
We present two different (although, related) solutions to this problem.
The source code for this solution is available on GitHub, and this README.md
is in: https://github.com/aerospike/url-tracker
##Installing and Running This Application
This is a Java application that uses Aerospike Java Client (version 3.0.27 or later), eclipse and Maven.
The pom.xml file shows Maven how to build the package.
Maven can easily be obtained from: http://maven.apache.org/
One set of steps for obtaining, building and running this application would be the following:
- get the package from github:
git clone [email protected]:aerospike/url-tracker.git - Import the package into eclipse as a git project
- Set the options (see options below) to point to a running Aerospike server
- Use the default commands.json file, or create your own.
##The Example Application Invocation
The application is a Java application, built as a Maven Project. It can be run
inside eclipse (a common choice while developing) or as a stand-alone program using
the generated jar file. The main class is UrlTracker and the main function is main().
In the eclipse tree -- the generated jar file is: url-tracker-new-1.0.0-full.jar
Also included in the package is the runapp file, which shows how to run the jar file directly:
java -cp target/url-tracker-new-1.0.0-full.jar com.aerospike.examples.ldt.UrlTracker
The application can be started with the following options:
options:
-c,--customer <arg> Generated Number of Customer Sets (default: 10)
-C,--Clean <arg> CLEAN all records at start of run (0==no, 1==yes) (default: 1)
-d,--debug Turn on DEBUG level prints.
-D,--CleanDuration <arg> Total seconds to run clean threads (default: 3600 sec)
-f,--filename <arg> Input File (default: commands.json)
-g,--generate Generate input data, rather than use Input File (default: false)
-h,--host <arg> Server hostname (default: localhost)
-I,--CleanInterval <arg> Time to sleep in seconds between cleaning (default: 1200 sec)
-L,--TimeToLive <arg> Number of seconds for a Site Visit Object to live (default: 600)
-M,--CleanMethod <arg> Method for cleaning expired values: 1:client, 2:UDF(default)
-n,--namespace <arg> Namespace (default: test)
-N,--NoLoad Do NOT load Data: Run ONLY the SiteVisit Updates and Clean
-O,--LoadOnly Load the Data, but do NOT update SiteVisits or Clean
-p,--port <arg> Server port (default: 3000)
-r,--records <arg> Generated number of Users per customer (default: 20)
-R,--Remove <arg> REMOVE all records at END of run (0==no, 1==yes) (default: 1)
-s,--set <arg> Set (default: demo)
-t,--type <arg> LDT Type (LMAP or LLIST) (default: LLIST)
-T,--Threads <arg> Number of threads to use in Generate Mode (default: 1)
-u,--usage Print usage.
-v,--visits <arg> Average number of SiteVisits per UserRecord (default: 1000)
By default, the application will expect to connect to an Aerospike Server running locally (localhost), at port 3000, using namespace "test" and set "demo", reading data from a local file called "commands.json" and using Large Ordered List (LLIST) to manage the site-visit collection.
#Program Modes
There are two modes in which the application can: "Generate Mode" and "JSON Data Mode".
"JSON Data Mode" uses a JSON file to feed commands to the URL-Tracker interpreter.
The commands create records, query records and delete records (explained below).
"Generate Mode" creates Customer Records, User Records and Site-Visit records according
to a pattern that is specified by the input parameters.
The "JSON Data Mode" is the default mode, and if there is no override to generate data
(with the -g option), the program will look for a commands.json file. There is an
example file shipped with this package, or a different commands file can be created.
##Data Generate Mode When the application is driven by a data generator, it uses the following pattern:
- To start in a pristine state, all Customer and User Records are removed
- All Customer Sets and customer records (one record per set) are populated
- For each Customer Set, all User Records are populated
- In a Pseudo-random pattern, Site-Visit records are generated for user records.
- At periodic intervals, a Customer Set will be scanned and EVERY User's record will be scrubbed of old site-visit records.
Example Invocations, with explanation:
##Invoke the URL-Tracker in generate mode using LLIST
./runapp -h Node_1 -n test -g -c 20 -r 200 -v 40000 -T 8 -M 2 -I 300 -D 84000 -L 600 -t LLIST
- -h Node_1 : Connect to a cluster where a node in that cluster is named "Node_1"
- -n test : Use the namespace named "test"
- -g : Use Generate Mode (as opposed to the JSON File Command mode)
- -c 20 : Create 20 Customer Sets (and a customer record in each set)
- -r 200 : Create 200 UserRecords in each set
- -v 40000 : Generate 40,000 Site Visit objects in each UserRecord of each Customer Set
- -T 8 : Use 8 parallel threads to write the Site Visit Objects
- -M 2 : Use the UDF mechanism to find and expire the old SiteVisit data objects
- -I 300 : Put the clean threads to sleep for 300 seconds between cleaning phases
- -D 84000 : Run the clean threads for a total duration of 84,000 seconds
- -L 600 : Set the Time To Live for each Site Visit Object to 600 seconds.
- -t LLIST : Use the LLIST LDT to hold the Site Visit Data in each User Record
##JSON Data Mode When the application is data driven (data read from the input file), the data file has the following format:
{
"command_file": "data file name",
"commands" : [ {"command element"}, {"command element"} ]
}
It is just a command_file object with a name and an array of commands.
There are several commands ("command element" instances) that are used for inserting, querying, expiring and removing data:
- Create New Customer ("new_customer")
- Create New User ("new_user")
- Append a new Site-Visit ("new_site_visit")
- Query User ("query_user"): Show all site-visit data for a user.
- Query Customer Set ("query_set"): Show all records in a customer set.
- Remove Expired Site Visit Entries ("remove_expired"): For a given expiration value, remove all site visit data that is older than the expiration value
- Remove Record ("remove_record"): Remove a record based on a record key.
- Remove All Records ("remove_all_records"): Remove all records from a given customer set.
This file represents the current example commands.json file that is found in this package.
{
"command_file": "DataFile_1.2",
"commands": [
{ "command": "new_customer",
"customer": {
"customer_id": "Indigo",
"contact": "Mr. Raj",
"set_name": "CustSetOne"
}
},
{
"command": "new_user",
"set_name": "CustSetOne" ,
"user": {
"name": "Bob",
"email": "[email protected]",
"phone": "(408) 555-1234",
"address": "1313 Mockingbird Lane",
"company": "aerospike"
}
},
{
"command": "new_user",
"set_name": "CustSetOne" ,
"user": {
"name": "Sue",
"email": "[email protected]",
"phone": "(408) 555-1234",
"address": "1313 Mockingbird Lane",
"company": "aerospike"
}
},
{
"command": "new_user",
"set_name": "CustSetOne" ,
"user": {
"name": "Joe",
"email": "[email protected]",
"phone": "(408) 555-1234",
"address": "1313 Mockingbird Lane",
"company": "aerospike"
}
},
{
"command": "new_user",
"set_name": "CustSetOne" ,
"user": {
"name": "Rick",
"email": "[email protected]",
"phone": "(408) 555-1234",
"address": "1313 Mockingbird Lane",
"company": "aerospike"
}
},
{
"command": "new_site_visit",
"set_name": "CustSetOne" ,
"user_name": "Bob",
"visit_info": {
"url": "www.aerospike.com",
"referrer": "xyz",
"page_title": "abc",
"ip_address": "1.2.3.4",
"date": 5000,
"expire": 6000
}
},
{
"command": "new_site_visit",
"set_name": "CustSetOne" ,
"user_name": "Sue",
"visit_info": {
"url": "www.aerospike.com",
"referrer": "xyz",
"page_title": "abc",
"ip_address": "1.2.3.4",
"date": 5001,
"expire": 6001
}
},
{
"command": "new_site_visit",
"set_name": "CustSetOne" ,
"user_name": "Sue",
"visit_info": {
"url": "www.microsoft.com",
"referrer": "xyz",
"page_title": "1a2b3c",
"ip_address": "1.2.3.4",
"date": 5010,
"expire": 6010
}
},
{
"command": "new_site_visit",
"set_name": "CustSetOne" ,
"user_name": "Bob",
"visit_info": {
"url": "www.google.com",
"referrer": "xyz",
"page_title": "abc",
"ip_address": "1.2.3.4",
"date": 5020,
"expire": 6020
}
},
{
"command": "new_site_visit",
"set_name": "CustSetOne" ,
"user_name": "Rick",
"visit_info": {
"url": "www.aerospike.com",
"referrer": "xyz",
"page_title": "abc",
"ip_address": "1.2.3.4",
"date": 5000,
"expire": 6000
}
},
{
"command": "new_site_visit",
"set_name": "CustSetOne" ,
"user_name": "Bob",
"visit_info": {
"url": "www.aerospike.com/documentation",
"referrer": "xyz",
"page_title": "abc",
"ip_address": "1.2.3.4",
"date": 5030,
"expire": 6030
}
},
{
"command": "query_user",
"set_name": "CustSetOne" ,
"user_name": "Bob"
},
{
"command": "remove_expired",
"set_name": "CustSetOne" ,
"user_name": "Bob",
"expire": 5500
},
{
"command": "query_set",
"set_name": "CustSetOne"
},
]
}