amazon web services - DynamoDB Schema Design -


i'm thinking of using amazon aws dynamodb project i'm working on. here's gist of situation:

i'm going gathering ton of energy usage data hundreds of machines (energy readings taken around every 5 minutes). each machine in zone, , each zone in network.

i'm going roll these individual readings zone , network, hour , day.

my thinking doing this, i'll able perform 1 query against dynamodb on network_day table, , return energy usage given day quickly.

here's schema @ point:

table_name      | hash_key   | range_key  | attributes ______________________________________________________ machine_reading | machine.id | epoch      | energy_use machine_hour    | machine.id | epoch_hour | energy_use machine_day     | machine.id | epoch_day  | energy_use zone_hour       | machine.id | epoch_hour | energy_use zone_day        | machine.id | epoch_day  | energy_use network_hour    | machine.id | epoch_hour | energy_use network_day     | machine.id | epoch_day  | energy_use

i'm not seeing great of performance in tests when run rollup cronjob, i'm wondering if more experience comment on key design? experience have far rds, i'm trying learn dynamodb.

edit:

basic structure cronjob i'm using rollups:

foreach network   foreach zone     foreach machine       add_unprocessed_readings_to_dynamo()       roll_up_fixture_hours_to_dynamo()       roll_up_fixture_days_to_dynamo()     end     roll_up_zone_hours_to_dynamo()     roll_up_zone_days_to_dynamo()   end   roll_up_network_hours_to_dynamo()   roll_up_network_days_to_dynamo() end

i use previous function's values in dynamo next roll up, i.e.

  • i use zone hours roll zone days
  • i use zone days roll network days

this (i think) causing lot of unnecessary reads/writes. right can manage low throughputs because sample size 100 readings. concerns begin when scales expected contain around 9,000,000 readings.

first things first, time series data in dynamodb hard right, not impossible.

dynamodb uses hash key shard data using machine.id means of going have hot keys. however, function of amount of data , expect iops be. dynamodb doesn't create 2nd shard until push past 1000 read or write iops. if expect below level may fine, if expect scale beyond may want redesign, include date component in hash key break things up.

regarding performance, hitting provisioned read or write throughput level? if raise them crazy high level , re-run test until bottleneck becomes code. simple setting throughput level appropriately.

however, regarding actual code, without seeing actual dynamodb queries performing possible issue reading data. make sure not reading more data need dynamodb. since range key date field use range conditional (not filter) reduce number of records need read.

make sure code executes rollup using multiple threads. if not able saturate dynamodb provisioned capacity issue may not dynamodb, may code. performing rollups using multiple threads in parallel should able see performance gains.


Comments

Post a Comment

Popular posts from this blog

powershell Start-Process exit code -1073741502 when used with Credential from a windows service environment -

i'm running strange behavior powershell start-process call. here call: $process = start-process ` "c:\somepath\mybinary.exe" ` -passthru ` -credential $defaultcredential ` -wait ` -workingdirectory "c:\somepath" ` -loaduserprofile if ($process.exitcode -ne 0) { #do } this call return exit code - 1073741502 . after quick search, exit code seems related generic error when program not load required dll (aka. status_dll_init_failed ). when run without -credential $credential program runs correctly. in order isolate problem, manually launched some.exe in prompt target credential , runs smoothly. so problem seems come way start-process cmdlet launch process. i found potential solutions problem tried apply no luck : link , link . would have idea of what's going on here ? edit 1: run proc mon monitoring program activities when launched directly or via powershell script. problem seems occur when loading kernelbase
Read more

twig - Using Twigbridge in a Laravel 5.1 Package -

i'm trying create laravel 5.1 package uses twigbridge instead of .php views. can normal .php views display fine, when try make use of .twig templates i'm met error. possible use twigbridge templates when creating views laravel package or pretty stuck blade , php ? edit 1: loading twig template without extending works. it's when start using {% extends 'file' %} error comes out. i error error loading /virtual/laravel/packages/username/mypackage/src/views/sample.twig: template "templates.master" not defined () in "/virtual/laravel/packages/username/mypackage/src/views/sample.twig" @ line 1. my folders username |_ mypackage |_ src |_ controllers - samplecontroller.php |_ views |_ templates - master.twig - sample.twig twig template sample.twig {% extends 'templates.master' %} {% block content %} {{ parent()
Read more

c# - LINQ join Entities from HashSet's, Join vs Dictionary vs HashSet performance -

i have hashset's each stores t, have written test application compares different relation algorithms can think of, i'm not pleased results getting. does there exist more efficient ways of achieving object relations, once testing? using system; using system.collections.generic; using system.linq; using system.text; using system.threading.tasks; using system.diagnostics; namespace linqtests { class program { static void main(string[] args) { hashset<user> usertable = new hashset<user>(); hashset<userproperty> userpropertytable = new hashset<userproperty>(); #region lets create dummy data. console.writeline("please wait while creating dummy data, can take while..."); console.writeline(""); int rows = 1000000; for(int x = 0; x < rows; x++) { random rnd = new random(); // ad
Read more