Sampling blocks - Bloklake

To chart a block-level metric over months, you don’t need every block. blockInterval returns every n-th block. With the Python client:

from bloklake import Bloklake

bl = Bloklake(chain="eth")

# Ethereum: 300 blocks ≈ 1 hour (12 s blocks)
page = bl.collect({
    "from_block": 19426000,
    "to_block": 19427499,
    "block_interval": 300,
    "fields": {"block": ["number", "timestamp", "base_fee_per_gas", "gas_used"]},
})
for b in page.blocks:
    print(b["number"], b["timestamp"], int(b["base_fee_per_gas"]) / 1e9, "gwei", b["gas_used"])

19426200 1710333467 66.415974755 gwei 15033090
19426500 1710337091 71.351142843 gwei 14971784
19426800 1710340715 53.076298262 gwei 29999024
19427100 1710344423 72.584137839 gwei 6740880
19427400 1710348107 74.480257029 gwei 16078401

Blocks whose number is a multiple of 300 (blockOffset defaults to 0). Use blockOffset to shift the phase: "blockInterval": 300, "blockOffset": 150 returns 19426150, 19426450, ….

Intervals for common periods

Period	Ethereum (12 s)	Robinhood Chain (~100 ms)
1 minute	5	600
1 hour	300	36,000
1 day	7,200	864,000

Block times aren’t perfectly regular, so these give approximately evenly spaced samples. Use timestamp for the x-axis, not the block number.

Cost

One year of Ethereum is about 2.6 million blocks. Hourly samples are about 8,760 rows; with the four fields above (about 40 bytes per row), that’s 350 KB. A year of hourly base fees costs almost nothing. The range still pages at 50,000 blocks, so a year is about 53 requests. Use the pagination loop.

Rules

blockInterval applies only to the blocks stream. It needs block fields selected.
Logs and transactions in the same request are not sampled.
With a logs or transactions filter, the blocks stream already holds only blocks with a match; sampling applies on top of that.