Poor performance with timeseries

Working with Data
,
1

I try to use timeseries, I have more than 1771200 documents in this collection.
My cluster is in version 6.0.14

new Schema(
  {
    analysis_area: {
      type: Types.String,
      required: true,
    },
    timestamp: {
      type: Types.Date,
      required: true,
    },
    waiting_time: { type: Types.Number },
  },
  {
    timeseries: {
      timeField: "timestamp",
      metaField: "analysis_area",
      granularity: "minutes",
    },
  }
).index({ analysis_area: 1, timestamp: 1 });

I have 4 queries :

  1. Select documents for an area and over 5 minutes.
  2. Select documents for an area and over 1 hour.
  3. Select documents for an area and over 1 day.
  4. Select documents for an area and over 1 month.

For queries 1 and 2, the execution times are correct, around 21ms for 55 documents and 70ms for 611 documents.

However, for queries 3 and 4, the execution times are significantly longer, around 870 ms for 28,789 documents and 12,200 ms for 417,706 documents.

Do you have any idea why performance is degraded so much for the last two queries and how to improve these performances?