static void Main(string[] args)
{
// Poll performance counter 4 times a second
TimeSpan pollingInterval = TimeSpan.FromSeconds(0.25);
// Take the total processor utilization performance counter
string categoryName = "Processor";
string counterName = "% Processor Time";
string instanceName = "_Total";
// Create an observable that feeds the performance counter periodically
IObservable <PerformanceCounterSample> source = new PerformanceCounterObservable(categoryName, counterName, instanceName, pollingInterval);
// Load the observable as a stream in Trill
var inputStream =
source.Select(e => StreamEvent.CreateStart(e.StartTime.Ticks, e)) // Create an IObservable of StreamEvent<>
.ToStreamable(OnCompletedPolicy.Flush(), PeriodicPunctuationPolicy.Count(4)); // Create a streamable with a punctuation every 4 events
// The query
long windowSize = TimeSpan.FromSeconds(2).Ticks;
var query = inputStream.AlterEventDuration(windowSize).Average(e => e.Value);
// Egress results and write to console
query.ToStreamEventObservable().ForEachAsync(e => WriteEvent(e)).Wait();
}
public static void Main(string[] args)
{
// Poll performance counter 4 times a second
var pollingInterval = TimeSpan.FromSeconds(0.25);
// Take the total processor utilization performance counter
string categoryName = "Processor";
string counterName = "% Processor Time";
string instanceName = "_Total";
// Create an observable that feeds the performance counter periodically
IObservable <PerformanceCounterSample> source =
new PerformanceCounterObservable(categoryName, counterName, instanceName, pollingInterval);
// Load the observable as a stream in Trill, injecting a punctuation every second. Because we use
// FlushPolicy.FlushOnPunctuation, this will also flush the data every second.
var inputStream =
source.Select(e => StreamEvent.CreateStart(e.StartTime.Ticks, e))
.ToStreamable(
null,
FlushPolicy.FlushOnPunctuation,
PeriodicPunctuationPolicy.Time((ulong)TimeSpan.FromSeconds(1).Ticks));
// The query
long windowSize = TimeSpan.FromSeconds(2).Ticks;
var query = inputStream.AlterEventDuration(windowSize).Average(e => e.Value);
// Egress results and write to console
query.ToStreamEventObservable().ForEachAsync(e => WriteEvent(e)).Wait();
}
public Subscription(PerformanceCounterObservable observable, IObserver <PerformanceCounterSample> observer)
{
// create a new counter for this subscription
this.counter = observable.createCounter();
this.pollingInterval = observable.pollingInterval;
this.observer = observer;
// seed previous sample to support computation
this.previousSample = this.counter.NextSample();
// create a timer to support polling counter at an interval
this.timer = new Timer(Sample);
this.timer.Change(this.pollingInterval.Milliseconds, -1);
}
public static void Main(string[] args)
{
// Poll performance counter 4 times a second
var pollingInterval = TimeSpan.FromSeconds(0.25);
// Take the total processor utilization performance counter
string categoryName = "Processor";
string counterName = "% Processor Time";
string instanceName = "_Total";
// Create an observable that feeds the performance counter periodically
IObservable <PerformanceCounterSample> source =
new PerformanceCounterObservable(categoryName, counterName, instanceName, pollingInterval);
// Load the observable as a stream in Trill, injecting a punctuation every second. Because we use
// FlushPolicy.FlushOnPunctuation, this will also flush the data every second.
var inputStream =
source.Select(e => StreamEvent.CreateStart(e.StartTime.Ticks, e))
.ToStreamable(
null,
FlushPolicy.FlushOnPunctuation,
PeriodicPunctuationPolicy.Time((ulong)TimeSpan.FromSeconds(1).Ticks));
// Query 1: Aggregate query
long windowSize = TimeSpan.FromSeconds(2).Ticks;
var query1 = inputStream.TumblingWindowLifetime(windowSize).Average(e => e.Value);
// Query 2: look for pattern of [CPU < 10] --> [CPU >= 10]
var query2 = inputStream
.AlterEventDuration(TimeSpan.FromSeconds(10).Ticks)
.Detect(default(Tuple <float, float>), // register to store CPU value
p => p
.SingleElement(e => e.Value < 10, (ts, ev, reg) => new Tuple <float, float>(ev.Value, 0))
.SingleElement(e => e.Value >= 10, (ts, ev, reg) => new Tuple <float, float>(reg.Item1, ev.Value)));
// Query 3: look for pattern of [k increasing CPU values --> drop CPU], report max, k
int k = 2;
var query3 = inputStream
.AlterEventDuration(TimeSpan.FromSeconds(10).Ticks)
.Detect(default(Tuple <float, int>), // register to store CPU value, incr count
p => p
.KleenePlus(e =>
e.SingleElement((ev, reg) => reg == null || ev.Value > reg.Item1, (ts, ev, reg) => new Tuple <float, int>(ev.Value, reg == null ? 1 : reg.Item2 + 1)))
.SingleElement((ev, reg) => ev.Value <reg.Item1 && reg.Item2> k, (ts, ev, reg) => reg),
allowOverlappingInstances: false);
// Egress results and write to console
query3.ToStreamEventObservable().ForEachAsync(e => WriteEvent(e)).Wait();
}
