Explainer
What Is a Parquet File — and How Do You Open One on Windows?
A colleague sent you a file ending in .parquet. It has no icon, double-clicking does nothing,
and every search result tells you to install Python.
The short answer
Parquet is a columnar, compressed, typed file format for tabular data — the standard way data warehouses hand over a dataset. Windows cannot open one natively. DataTray's Data Explorer reads it directly, including a folder of partitioned parts, and converts to CSV or Excel with nothing else to install.
What makes it different from a CSV
A CSV stores your data row by row, as text, with no types and no compression. Parquet does three things differently, and all three follow from the first:
It stores columns together. All the values from one column are contiguous. A query that reads two columns out of forty touches only those two, which is why analytics tools are so much faster on it.
It compresses. Values within a column are similar to each other — dates near dates, repeated categories, a narrow range of numbers — which compresses far better than a row of mixed types. The same data is frequently a fraction of the CSV's size.
It records the schema. Column names and types live inside the file. Nothing has to guess whether a column is a date, and nothing loses your leading zeros on open.
Why you were sent one
Because whoever produced it is working in a pipeline where those properties matter, and Parquet is what their tooling emits by default — Spark, dbt, a warehouse export, an S3 data lake.
It is not an attempt to be awkward. It is just that a format optimised for machines reading columns is a poor fit for a person double-clicking a file, and the handoff between those two worlds is exactly where people get stuck.
The directory thing
A large Parquet dataset is often not a file at all but a folder full of files —
part-00000.parquet, part-00001.parquet, and so on, sometimes nested inside directories named
like region=West/.
That is one table, deliberately split. Do not try to open the parts individually: point a reader
at the directory, and it reads them as a single table. DataTray does this, and the
region=West/ folder names come through as a column.
Opening one without installing anything
- Open Data Explorer and drop in the
.parquetfile or its folder. - The rows appear, along with the schema Parquet already recorded — types included.
- Convert to
.xlsxor CSV if you need to send it to someone in a spreadsheet.
One caveat worth expecting: converting Parquet to CSV usually produces a much larger file, because you are discarding the compression and the types. If size matters, keep the Parquet and send a filtered extract instead.
When to send Parquet, and when not to
Send Parquet to someone working in a pipeline, or when the dataset is large and the recipient will query rather than read it.
Send .xlsx to someone who is going to look at it. Parquet's advantages are invisible to a
person opening a file, and its one disadvantage — nothing on their machine can read it — is very
visible indeed.
Every other free way to inspect a Parquet file asks you to upload it first, or to install a programming language. Neither is necessary to look at a table.
Do this in one pass with — Data Explorer
Read-only, always. Open a CSV, Excel workbook, Parquet directory or JSON file of any size and look at the rows — then switch to Columns for a full data profile: what every column actually holds, where the nulls and the mixed types and the outliers are, a data-quality read on each one, and a data dictionary you can hand to someone. A third tab takes SQL if you want it, and never asks you to write any if you don’t. Nothing here modifies the file; the only things written are the exports and reports you ask for. Learn more about the Data Explorer.
Windows 10/11 (x64) · Free · No account