![どれくらい簡単か
• C#の場合
namespace Microsoft.ServiceBus.Samples
{
using System.Diagnostics;
using System.Runtime.Serialization.Json;
using System.Threading;
using System;
using System.Collections.Generic;
using System.IO;
using System.Linq;
using System.Text;
using System.Threading.Tasks;
using Microsoft.ServiceBus.Messaging;
using Newtonsoft.Json;
public class SimpleEventProcessor : IEventProcessor
{
IDictionary<string, int> map;
PartitionContext partitionContext;
Stopwatch checkpointStopWatch;
public SimpleEventProcessor()
{
this.map = new Dictionary<string, int>();
}
public Task OpenAsync(PartitionContext context)
{
Console.WriteLine(string.Format("SimpleEventProcessor initialize. Partition: '{0}', Offset: '{1}'",
context.Lease.PartitionId, context.Lease.Offset));
this.partitionContext = context;
this.checkpointStopWatch = new Stopwatch();
this.checkpointStopWatch.Start();
return Task.FromResult<object>(null);
}
public async Task ProcessEventsAsync(PartitionContext context, IEnumerable<EventData> events)
{
try
{
foreach (EventData eventData in events)
{
int data;
var newData = this.DeserializeEventData(eventData);
string key = eventData.PartitionKey;
// Name of device generating the event acts as hash key to retrieve average computed for it so
far
if (!this.map.TryGetValue(key, out data))
{
// If this is the first time we got data for this device then generate new state
this.map.Add(key, -1);
}
// Update data
data = newData.Temperature;
this.map[key] = data;](https://image.slidesharecdn.com/20150905streamanalytics-150905142946-lva1-app6891/85/20150905-stream-analytics-8-320.jpg)
20150905 stream analytics
- 1. Stream Analyticsのさわり Go(5) JAZUG / JAZUG5周年総会 大田 一希
- 2. 自己紹介 • 名前/SNS • 大田 一希 • Twitter: @okazuki • Blog • かずきのBlog@hatena http://okazuki.hatenablog.com/ • その他 • Microsoft MVP for Windows Platform Development • Javaもやってるよ
- 5. Stream Analyticsとは • 流れるデータを処理する基盤 • IoTのような無慈悲なセンサーデータを さばくのに向いてると思う • EventHubやBlobのデータを処理で きる • SQLライクな言語で書ける
- 8. どれくらい簡単か • C#の場合 namespace Microsoft.ServiceBus.Samples { using System.Diagnostics; using System.Runtime.Serialization.Json; using System.Threading; using System; using System.Collections.Generic; using System.IO; using System.Linq; using System.Text; using System.Threading.Tasks; using Microsoft.ServiceBus.Messaging; using Newtonsoft.Json; public class SimpleEventProcessor : IEventProcessor { IDictionary<string, int> map; PartitionContext partitionContext; Stopwatch checkpointStopWatch; public SimpleEventProcessor() { this.map = new Dictionary<string, int>(); } public Task OpenAsync(PartitionContext context) { Console.WriteLine(string.Format("SimpleEventProcessor initialize. Partition: '{0}', Offset: '{1}'", context.Lease.PartitionId, context.Lease.Offset)); this.partitionContext = context; this.checkpointStopWatch = new Stopwatch(); this.checkpointStopWatch.Start(); return Task.FromResult<object>(null); } public async Task ProcessEventsAsync(PartitionContext context, IEnumerable<EventData> events) { try { foreach (EventData eventData in events) { int data; var newData = this.DeserializeEventData(eventData); string key = eventData.PartitionKey; // Name of device generating the event acts as hash key to retrieve average computed for it so far if (!this.map.TryGetValue(key, out data)) { // If this is the first time we got data for this device then generate new state this.map.Add(key, -1); } // Update data data = newData.Temperature; this.map[key] = data;
- 13. 時間を扱う • データのグルーピング（集計）のとき やJOINのときに使う • データを指定した時間でぶった切る • その中で集計する • 例えばWindow化関数 • TumblingWindow • HoppingWindow • SlidingWindow
- 14. TumblingWindow • 指定した時間でデータを区切る 5sec 10sec 15sec 20sec
- 15. TumblingWindow • SELECT COUNT(*) AS Cnt FROM Input GROUP BY TumblingWindow(second, 5) 5sec 10sec 15sec 20sec 3 3 3 3
- 16. まとめ • Stream Analytics • 大量データをさばくための基盤 • SQLライクな言語で処理を記述できる • 使い始め超簡単 • 時間という概念大事