Creating graph for Cloud Native Geo Conference 2025 to test async-tiff and rasterio

create_async_tiff_graph.py

import altair as alt
import pandas as pd

recv_bytes_per_second_mean = pd.DataFrame(
    {
        "Group": ["rasterio", "async-tiff"],
        "Throughput (bytes/s)": [972102.34, 24319246.71],
    },
)

recv_bytes_per_second_per_cpu_mean = pd.DataFrame(
    {
        "Group": ["rasterio", "async-tiff"],
        "Throughput per CPU second": [760349.7069, 11429263.13],
    },
)

memory_usage_bytes_mean = pd.DataFrame(
    {
        "Group": ["rasterio", "async-tiff"],
        "Memory Usage (MiB)": [91331281.77 / 1024 / 1024, 491729766.7 / 1024 / 1024],
    },
)

recv_bytes_per_second_mean_chart = (
    alt.Chart(recv_bytes_per_second_mean)
    .mark_bar()
    .encode(
        x=alt.X("Group:N", title="Library"),
        y=alt.Y("Throughput (bytes/s):Q"),
        color=alt.Color("Group:N", legend=alt.Legend(title="Library")),
    )
    .properties(
        width=150,
        height=200,
        title=alt.TitleParams(
            text="Throughput (higher is better)",
            fontSize=14,
        ),
    )
)

recv_bytes_per_second_per_cpu_mean_chart = (
    alt.Chart(recv_bytes_per_second_per_cpu_mean)
    .mark_bar()
    .encode(
        x=alt.X("Group:N", title="Library"),
        y=alt.Y("Throughput per CPU second:Q"),
        color=alt.Color("Group:N", legend=alt.Legend(title="Library")),
    )
    .properties(
        width=150,
        height=200,
        title=alt.TitleParams(
            text="Throughput/CPU sec (higher is better)",
            fontSize=14,
        ),
    )
)

memory_usage_bytes_mean_chart = (
    alt.Chart(memory_usage_bytes_mean)
    .mark_bar()
    .encode(
        x=alt.X("Group:N", title="Library"),
        y=alt.Y("Memory Usage (MiB):Q"),
        color=alt.Color("Group:N", legend=alt.Legend(title="Library")),
    )
    .properties(
        width=150,
        height=200,
        title=alt.TitleParams(text="Memory Usage (lower is better)", fontSize=14),
    )
)

final_chart = (
    alt.hconcat(
        recv_bytes_per_second_mean_chart,
        recv_bytes_per_second_per_cpu_mean_chart,
        memory_usage_bytes_mean_chart,
    )
    .properties(
        title="Comparing async-tiff and rasterio for opening & parsing COG header.",
    )
    .configure_title(anchor="middle", fontSize=18)
)

final_chart.save("async-tiff.png", scale_factor=2)

obstore_perf_graph.py

import altair as alt
import pandas as pd

recv_bytes_per_second_mean = pd.DataFrame(
    {
        "Group": ["obstore", "requests", "aioboto3", "aiohttp"],
        "Throughput (bytes/s)": [253384242.4, 81147299.37, 87910179.82, 231605561.1],
    },
)

recv_bytes_per_second_per_cpu_mean = pd.DataFrame(
    {
        "Group": ["obstore", "requests", "aioboto3", "aiohttp"],
        "Throughput per CPU second": [43811471.59, 8542193.1, 11903911.6, 38808248.24],
    },
)

memory_usage_bytes_mean = pd.DataFrame(
    {
        "Group": ["obstore", "requests", "aioboto3", "aiohttp"],
        "Memory Usage (MiB)": [
            1060266099 / 1024 / 1024,
            126520167.3 / 1024 / 1024,
            546641897.1 / 1024 / 1024,
            1642166646 / 1024 / 1024,
        ],
    },
)

recv_bytes_per_second_mean_chart = (
    alt.Chart(recv_bytes_per_second_mean)
    .mark_bar()
    .encode(
        x=alt.X("Group:N", title="Library"),
        y=alt.Y("Throughput (bytes/s):Q"),
        color=alt.Color("Group:N", legend=alt.Legend(title="Library")),
    )
    .properties(
        width=150,
        height=200,
        title=alt.TitleParams(
            text="Throughput (higher is better)",
            fontSize=14,
        ),
    )
)

recv_bytes_per_second_per_cpu_mean_chart = (
    alt.Chart(recv_bytes_per_second_per_cpu_mean)
    .mark_bar()
    .encode(
        x=alt.X("Group:N", title="Library"),
        y=alt.Y("Throughput per CPU second:Q"),
        color=alt.Color("Group:N", legend=alt.Legend(title="Library")),
    )
    .properties(
        width=150,
        height=200,
        title=alt.TitleParams(
            text="Throughput/CPU sec (higher is better)",
            fontSize=14,
        ),
    )
)

memory_usage_bytes_mean_chart = (
    alt.Chart(memory_usage_bytes_mean)
    .mark_bar()
    .encode(
        x=alt.X("Group:N", title="Library"),
        y=alt.Y("Memory Usage (MiB):Q"),
        color=alt.Color("Group:N", legend=alt.Legend(title="Library")),
    )
    .properties(
        width=150,
        height=200,
        title=alt.TitleParams(text="Memory Usage (lower is better)", fontSize=14),
    )
)

final_chart = (
    alt.hconcat(
        recv_bytes_per_second_mean_chart,
        recv_bytes_per_second_per_cpu_mean_chart,
        memory_usage_bytes_mean_chart,
    )
    .properties(
        title="Comparing libraries for requesting initial 32kb of file on S3",
    )
    .configure_title(anchor="middle", fontSize=18)
)

final_chart.save("obstore-concurrency.png", scale_factor=2)