First, a short hint: Kommentare im Quellcode sind gut. Und ich verstehe, warum man sie in seiner Muttersprache schreibt. Aber das kann es für andere schwieriger machen, zu helfen. Auch wenn automatische Übersetzer inzwischen ganz gut sind, kann man sie nicht einfach so auf Quellcode und Kommentare anwenden - geschweige denn die Beschriftungen der Knöpfe. Ich habe einfach mal den grünen Knopf gedrückt. Zum Glück hat das hier gereicht 
Back to the problem, starting with another small hint:
You don’t have to wait a few minutes when you just change the line
dataPushInterval = setInterval(simulateDataPush, 2000);
to
dataPushInterval = setInterval(simulateDataPush, 50);
Or even better: Pull that out into a variable
const simulatePushDelayMs = 50;
...
dataPushInterval = setInterval(simulateDataPush, simulatePushDelay);
so that it can quickly and easily be changed for tests.
Now, the actual problem:
You can see that the CPU usage quickly raises to 100%. On my machine, it reaches 100% when there are about ~700 points in each of these 10 polylines.
When you do a profiler run with this, then you’ll see where the time is spent:
This may not immediately give the answer, but it can be a hint: When you create a default polyline in CesiumJS, then it uses an ArcType.GEODESIC by default. This means that it “subdivides” this line, so that it follows the curvature of earth - for example:
Computing this curvature is expensive. Your polylines only cover a small region, and the curvature of earth does not matter at this scale.
So you could insert…
// Erzeuge die Entität :-)
const entity = dataSource.entities.add({
id: `entity_${i}`,
polyline: {
positions: positionsProperty,
width: 5,
material: new Cesium.PolylineOutlineMaterialProperty({
color: colors[i-1],
outlineWidth: 2,
outlineColor: Cesium.Color.WHITE
}),
clampToGround: false,
arcType: Cesium.ArcType.NONE // -------------------------- ! THIS LINE
}
});
… to set the arcType to NONE (meaning that all points are just connected with a straight line).
This way, on my machine, the CPU usage raises much slower, and only at ~1700 points per entity, it reaches 100%.
(There may be further room for optimizations. But regardless of how well you optimize it: There will always be a limit for the number of points. Maybe this improvement is already enough for your use-case…)