andygrove commented on code in PR #214:
URL: https://github.com/apache/datafusion-site/pull/214#discussion_r4432836228
##########
content/blog/2026-10-25-comet-nested-tpch.md:
##########
@@ -0,0 +1,225 @@
+---
+layout: post
+title: "Nested TPC-H: A Benchmark for Nested Data in Apache DataFusion Comet"
+date: 2026-10-25
+author: pmc
+categories: [performance, subprojects]
+---
+
+<!--
+{% comment %}
+Licensed to the Apache Software Foundation (ASF) under one or more
+contributor license agreements. See the NOTICE file distributed with
+this work for additional information regarding copyright ownership.
+The ASF licenses this file to you under the Apache License, Version 2.0
+(the "License"); you may not use this file except in compliance with
+the License. You may obtain a copy of the License at
+
+http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+{% endcomment %}
+-->
+
+[TOC]
+
+Apache DataFusion Comet tracked its performance with TPC-H and TPC-DS, and the
numbers were encouraging. Real user pipelines showed smaller gains, or none at
all. What those pipelines had in common was nested data, which neither
benchmark contains.
+
+To measure that gap, we built nested TPC-H. At scale factor 1000 (1 TB), the
upcoming Comet 1.2.0 runs all 22 of its queries in 790 s, against 1,519 s for
Spark 4.1.3.
+
+<img
+src="/blog/images/comet-nested-tpch/intro-total.svg"
+width="100%"
+class="img-fluid"
+alt="Total execution time at SF1000: Spark 4.1.3 1,519 s, Comet 1.1.0 869 s,
upcoming Comet 1.2.0 790 s"
+/>
+
+Nested TPC-H is a TPC-H derived benchmark in the Comet repository
([#7044](https://github.com/apache/datafusion-comet/pull/7044)). The TPC-H data
and the questions the queries answer are unchanged. The tables are reshaped to
store most columns in arrays of structs, and the queries are rewritten to
unnest those arrays. The data and queries are engine neutral, so they can be
used to compare Spark, Comet and other engines.
+
+The first runs showed where the gaps were: some queries ran slower on Comet
than on Spark, even though Comet accelerates the same queries on flat TPC-H.
+
+## Why: TPC-H is flat, real data is not
+
+[TPC-H](https://www.tpc.org/tpch/) is one of the most cited analytical
benchmarks, and its schema is flat: eight normalized tables with primitive
columns. Its 22 queries exercise well-understood choke points such as
aggregation, join ordering and correlated subqueries (Boncz, Neumann and
Erling, ["TPC-H Analyzed"](https://doi.org/10.1007/978-3-319-04936-6_5), TPCTC
2013).
+
+Real data is often nested. Event logs, JSON, Avro and Protobuf records, and
many data lake tables are full of structs and arrays, which Parquet stores with
the record shredding scheme from Dremel (Melnik et al.,
["Dremel"](http://www.vldb.org/pvldb/vldb2010/papers/R29.pdf), VLDB 2010).
Nested data adds work that flat benchmarks never exercise:
+
+- **Reading.** Rebuilding list offsets and validity bitmaps from repetition
and definition levels, and decoding only the fields a query uses.
+- **Planning.** Nested column pruning and predicate pushdown through `explode`.
+- **Operators.** `explode`, field access, and sort, join and exchange over
wide variable-size values.
+- **Memory and spilling.** Variable-size values make memory use and spill
volume harder to predict.
+- **Conversion.** Moving nested values between a JVM and a native
representation.
+
+Comet hit the shuffle case in
[#5397](https://github.com/apache/datafusion-comet/issues/5397), where
round-robin shuffle writes of wide nested fields were two to three times slower
than Spark's. The issue notes that it was "invisible for TPC-* benchmarks
because of historical flat schema".
+
+We needed a benchmark that filled the gap between academic benchmarking and
real user workloads.
+
+## How: an adapter that nests TPC-H
+
+We did not want to invent a new benchmark from scratch. TPC-H is well
understood, widely published and easy to generate, and its business shape
(customers, orders, line items, suppliers and parts) is what makes its results
meaningful. The goal was to preserve that shape and those questions while
making every query process nested data.
+
+The idea turned out to be surprisingly simple: a transformation of both the
data and the queries that leaves the key and partition columns at the top level
and nests everything else. The partition columns are `p_brand`, `c_mktsegment`,
`o_orderdate` and `l_shipdate`, and every other non-key column moves into one
`array<struct>` named `<table>_data`. Each array holds exactly one element, so
exploding it never changes the row count and the queries return the same
answers as TPC-H. For `lineitem`:
+
+```
+-- flat -- nested
+l_orderkey BIGINT l_orderkey BIGINT
+l_partkey BIGINT l_partkey BIGINT
+l_suppkey BIGINT l_suppkey BIGINT
+l_linenumber INT l_shipdate DATE -- partition
column
+l_quantity DECIMAL(12,2) lineitem_data ARRAY<STRUCT<
+l_extendedprice DECIMAL(12,2) l_linenumber: INT,
+l_discount DECIMAL(12,2) l_quantity: DECIMAL(12,2),
+l_tax DECIMAL(12,2) l_extendedprice: DECIMAL(12,2),
+l_returnflag STRING l_discount: DECIMAL(12,2),
+l_linestatus STRING l_tax: DECIMAL(12,2),
+l_shipdate DATE l_returnflag: STRING,
+l_commitdate DATE l_linestatus: STRING,
+l_receiptdate DATE l_commitdate: DATE,
+l_shipinstruct STRING l_receiptdate: DATE,
+l_shipmode STRING l_shipinstruct: STRING,
+l_comment STRING l_shipmode: STRING,
+ l_comment: STRING>>
+```
+
+Each of the 22 queries is rewritten to unnest only the arrays it needs. The
queries are in
[`benchmarks/nested/queries`](https://github.com/apache/datafusion-comet/tree/main/benchmarks/nested/queries).
+
+```sql
+from (select l_orderkey, l_partkey, l_suppkey, l_shipdate,
+ explode(lineitem_data) as lineitem_data from lineitem) as lineitem
+where lineitem_data.l_quantity < 24
+```
+
+[`nest_tpch.scala`](https://github.com/apache/datafusion-comet/blob/main/benchmarks/nested/scripts/nest_tpch.scala)
converts an existing flat TPC-H dataset to the nested layout. The source and
the target can each be Parquet or Iceberg. The script runs under `spark-shell`
on a laptop or a cluster, casts columns to fixed types, writes zstd-compressed
Parquet with 512 MiB row groups, partitions by the columns above and checks row
counts:
+
+```bash
+spark-shell -i scripts/nest_tpch.scala \
+ --conf spark.nesttpch.source=parquet:/data/tpch/sf10 \
+ --conf spark.nesttpch.target=parquet:/data/tpch-nested/sf10
+```
+
+The
[README](https://github.com/apache/datafusion-comet/tree/main/benchmarks/nested)
has the Iceberg variants and all settings.
+
+The script and queries in the Comet repository also serve as an example. You
can customize them to test any nested complexity, such as multi-element arrays,
maps or deeper nesting.
+
+## Benchmark setup
+
+- **Data**: nested TPC-H at scale factor 1000 (1 TB), created from flat TPC-H
with `nest_tpch.scala`.
+- **Tables**: Apache Iceberg tables in a Hadoop catalog on S3, with
zstd-compressed Parquet data files, 512 MiB row groups and the partitioning
above. Comet reads them with its native Iceberg scan, which is built on
[iceberg-rust](https://github.com/apache/iceberg-rust).
+- **Versions**: Spark 4.1.3 (Scala 2.13, JDK 17) and Iceberg 1.11.0. Comet
1.1.0, and the upcoming Comet 1.2.0 built from `branch-1.2` at `746246049`. The
`1.2.0-rc1` tag is five commits ahead of that build, all backported bug fixes
that we don't expect to change these numbers. Both Comet builds use the native
Iceberg scan and the default allocator (glibc).
+- **Cluster**: Kubernetes, with 32 executors of 16 task slots each (8 CPUs per
executor pod), shuffle and spill files on a block volume attached to each
executor, and `spark.sql.shuffle.partitions=512`.
+- **Method**: the queries in `benchmarks/nested`, run from `spark-shell`. Each
query runs three times in one application with no warm-up, timed from
`spark.sql()` to the end of `collect()`. Figures are the fastest iteration, and
totals are sums of the per-query times. Every query returned the expected TPC-H
row count.
+- **Memory**: Spark runs with a 64 GiB heap and no off-heap memory. Comet runs
with a 32 GiB heap and 64 GiB of off-heap memory, which gives each task 4 GiB
of native memory. Both add 10 GiB of memory overhead.
+
+Both engines share these settings: G1 GC, Kryo,
`spark.io.compression.codec=zstd`, `spark.memory.fraction=0.6`,
`spark.memory.storageFraction=0.2`, `spark.sql.sources.ignoreDataLocality=true`
and `spark.executor.memoryOverhead=10g`. Comet adds:
+
+```
+spark.executor.memory=32g
+spark.memory.offHeap.enabled=true
+spark.memory.offHeap.size=64g
+spark.plugins=org.apache.spark.CometPlugin
+spark.shuffle.manager=org.apache.spark.sql.comet.execution.shuffle.CometShuffleManager
+spark.comet.parquet.prefetchNextFileMetadata.enabled=true
+spark.comet.scan.icebergNative.enabled=true
+spark.comet.scan.icebergNative.dataFileConcurrencyLimit=4
+spark.executor.defaultJavaOptions=-XX:TrimNativeHeapInterval=5000
-Xlog:trimnative
+```
+
+Two of these settings were needed to run the full suite on Comet:
+
+- **Off-heap memory.** With a 64 GiB heap and 32 GiB of off-heap memory, each
task gets 2 GiB of native memory. That is not enough for q21's sort-merge join
stage on Comet 1.2.0, which allocates about 2.2 GiB per task and reserves up to
3.1 GiB, so tasks failed and were retried.
+- **`-XX:TrimNativeHeapInterval`.** Without it, glibc keeps native memory that
Comet has freed, and executor resident memory grows from one query to the next.
Late in the suite, q21 had 11 executors OOMKilled while Comet's memory pools
were well inside their limits. Trimming every 5 s kept executors at 61 to 70
GiB of resident memory and cost about 4% on the other queries. The Comet tuning
guide now recommends it
([#7342](https://github.com/apache/datafusion-comet/pull/7342)).
+
+## What nested benchmarks revealed
+
+The nested schema uses single-element arrays of structs, with no maps for now.
Even that was enough to expose issues that were invisible on flat TPC-H. In our
first runs, Comet was slower than Spark on q20 and q21, and barely faster on
q02. Tracing those queries surfaced a lot of technical debt in how nested data
is read, shuffled and spilled, in Comet and in the iceberg-rust reader
underneath it.
Review Comment:
The "first runs" claims don't line up with the results table. Here q20 and
q21 are slower than Spark and q02 is barely faster. Line 143 says q20 "was no
faster", and line 43 says "some queries ran slower on Comet". In the table, the
only query where 1.1.0 is slower than Spark is q14, and q20 and q21 are 1.39x
and 1.52x on 1.1.0. Readers will try to reconcile these. Could we say which
build and configuration the first runs used (e.g. an earlier build, before the
fixes below), or drop the per-query claims? Line 208 already explains this for
the q21 380 s to 250 s numbers, so the same qualifier could be reused here.
##########
content/blog/2026-10-25-comet-nested-tpch.md:
##########
@@ -0,0 +1,225 @@
+---
+layout: post
+title: "Nested TPC-H: A Benchmark for Nested Data in Apache DataFusion Comet"
+date: 2026-10-25
+author: pmc
+categories: [performance, subprojects]
+---
+
+<!--
+{% comment %}
+Licensed to the Apache Software Foundation (ASF) under one or more
+contributor license agreements. See the NOTICE file distributed with
+this work for additional information regarding copyright ownership.
+The ASF licenses this file to you under the Apache License, Version 2.0
+(the "License"); you may not use this file except in compliance with
+the License. You may obtain a copy of the License at
+
+http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+{% endcomment %}
+-->
+
+[TOC]
+
+Apache DataFusion Comet tracked its performance with TPC-H and TPC-DS, and the
numbers were encouraging. Real user pipelines showed smaller gains, or none at
all. What those pipelines had in common was nested data, which neither
benchmark contains.
+
+To measure that gap, we built nested TPC-H. At scale factor 1000 (1 TB), the
upcoming Comet 1.2.0 runs all 22 of its queries in 790 s, against 1,519 s for
Spark 4.1.3.
+
+<img
+src="/blog/images/comet-nested-tpch/intro-total.svg"
+width="100%"
+class="img-fluid"
+alt="Total execution time at SF1000: Spark 4.1.3 1,519 s, Comet 1.1.0 869 s,
upcoming Comet 1.2.0 790 s"
+/>
+
+Nested TPC-H is a TPC-H derived benchmark in the Comet repository
([#7044](https://github.com/apache/datafusion-comet/pull/7044)). The TPC-H data
and the questions the queries answer are unchanged. The tables are reshaped to
store most columns in arrays of structs, and the queries are rewritten to
unnest those arrays. The data and queries are engine neutral, so they can be
used to compare Spark, Comet and other engines.
+
+The first runs showed where the gaps were: some queries ran slower on Comet
than on Spark, even though Comet accelerates the same queries on flat TPC-H.
+
+## Why: TPC-H is flat, real data is not
+
+[TPC-H](https://www.tpc.org/tpch/) is one of the most cited analytical
benchmarks, and its schema is flat: eight normalized tables with primitive
columns. Its 22 queries exercise well-understood choke points such as
aggregation, join ordering and correlated subqueries (Boncz, Neumann and
Erling, ["TPC-H Analyzed"](https://doi.org/10.1007/978-3-319-04936-6_5), TPCTC
2013).
+
+Real data is often nested. Event logs, JSON, Avro and Protobuf records, and
many data lake tables are full of structs and arrays, which Parquet stores with
the record shredding scheme from Dremel (Melnik et al.,
["Dremel"](http://www.vldb.org/pvldb/vldb2010/papers/R29.pdf), VLDB 2010).
Nested data adds work that flat benchmarks never exercise:
+
+- **Reading.** Rebuilding list offsets and validity bitmaps from repetition
and definition levels, and decoding only the fields a query uses.
+- **Planning.** Nested column pruning and predicate pushdown through `explode`.
+- **Operators.** `explode`, field access, and sort, join and exchange over
wide variable-size values.
+- **Memory and spilling.** Variable-size values make memory use and spill
volume harder to predict.
+- **Conversion.** Moving nested values between a JVM and a native
representation.
+
+Comet hit the shuffle case in
[#5397](https://github.com/apache/datafusion-comet/issues/5397), where
round-robin shuffle writes of wide nested fields were two to three times slower
than Spark's. The issue notes that it was "invisible for TPC-* benchmarks
because of historical flat schema".
+
+We needed a benchmark that filled the gap between academic benchmarking and
real user workloads.
+
+## How: an adapter that nests TPC-H
+
+We did not want to invent a new benchmark from scratch. TPC-H is well
understood, widely published and easy to generate, and its business shape
(customers, orders, line items, suppliers and parts) is what makes its results
meaningful. The goal was to preserve that shape and those questions while
making every query process nested data.
+
+The idea turned out to be surprisingly simple: a transformation of both the
data and the queries that leaves the key and partition columns at the top level
and nests everything else. The partition columns are `p_brand`, `c_mktsegment`,
`o_orderdate` and `l_shipdate`, and every other non-key column moves into one
`array<struct>` named `<table>_data`. Each array holds exactly one element, so
exploding it never changes the row count and the queries return the same
answers as TPC-H. For `lineitem`:
+
+```
+-- flat -- nested
+l_orderkey BIGINT l_orderkey BIGINT
+l_partkey BIGINT l_partkey BIGINT
+l_suppkey BIGINT l_suppkey BIGINT
+l_linenumber INT l_shipdate DATE -- partition
column
+l_quantity DECIMAL(12,2) lineitem_data ARRAY<STRUCT<
+l_extendedprice DECIMAL(12,2) l_linenumber: INT,
+l_discount DECIMAL(12,2) l_quantity: DECIMAL(12,2),
+l_tax DECIMAL(12,2) l_extendedprice: DECIMAL(12,2),
+l_returnflag STRING l_discount: DECIMAL(12,2),
+l_linestatus STRING l_tax: DECIMAL(12,2),
+l_shipdate DATE l_returnflag: STRING,
+l_commitdate DATE l_linestatus: STRING,
+l_receiptdate DATE l_commitdate: DATE,
+l_shipinstruct STRING l_receiptdate: DATE,
+l_shipmode STRING l_shipinstruct: STRING,
+l_comment STRING l_shipmode: STRING,
+ l_comment: STRING>>
+```
+
+Each of the 22 queries is rewritten to unnest only the arrays it needs. The
queries are in
[`benchmarks/nested/queries`](https://github.com/apache/datafusion-comet/tree/main/benchmarks/nested/queries).
+
+```sql
+from (select l_orderkey, l_partkey, l_suppkey, l_shipdate,
+ explode(lineitem_data) as lineitem_data from lineitem) as lineitem
+where lineitem_data.l_quantity < 24
+```
+
+[`nest_tpch.scala`](https://github.com/apache/datafusion-comet/blob/main/benchmarks/nested/scripts/nest_tpch.scala)
converts an existing flat TPC-H dataset to the nested layout. The source and
the target can each be Parquet or Iceberg. The script runs under `spark-shell`
on a laptop or a cluster, casts columns to fixed types, writes zstd-compressed
Parquet with 512 MiB row groups, partitions by the columns above and checks row
counts:
+
+```bash
+spark-shell -i scripts/nest_tpch.scala \
+ --conf spark.nesttpch.source=parquet:/data/tpch/sf10 \
+ --conf spark.nesttpch.target=parquet:/data/tpch-nested/sf10
+```
+
+The
[README](https://github.com/apache/datafusion-comet/tree/main/benchmarks/nested)
has the Iceberg variants and all settings.
+
+The script and queries in the Comet repository also serve as an example. You
can customize them to test any nested complexity, such as multi-element arrays,
maps or deeper nesting.
+
+## Benchmark setup
+
+- **Data**: nested TPC-H at scale factor 1000 (1 TB), created from flat TPC-H
with `nest_tpch.scala`.
+- **Tables**: Apache Iceberg tables in a Hadoop catalog on S3, with
zstd-compressed Parquet data files, 512 MiB row groups and the partitioning
above. Comet reads them with its native Iceberg scan, which is built on
[iceberg-rust](https://github.com/apache/iceberg-rust).
+- **Versions**: Spark 4.1.3 (Scala 2.13, JDK 17) and Iceberg 1.11.0. Comet
1.1.0, and the upcoming Comet 1.2.0 built from `branch-1.2` at `746246049`. The
`1.2.0-rc1` tag is five commits ahead of that build, all backported bug fixes
that we don't expect to change these numbers. Both Comet builds use the native
Iceberg scan and the default allocator (glibc).
+- **Cluster**: Kubernetes, with 32 executors of 16 task slots each (8 CPUs per
executor pod), shuffle and spill files on a block volume attached to each
executor, and `spark.sql.shuffle.partitions=512`.
+- **Method**: the queries in `benchmarks/nested`, run from `spark-shell`. Each
query runs three times in one application with no warm-up, timed from
`spark.sql()` to the end of `collect()`. Figures are the fastest iteration, and
totals are sums of the per-query times. Every query returned the expected TPC-H
row count.
+- **Memory**: Spark runs with a 64 GiB heap and no off-heap memory. Comet runs
with a 32 GiB heap and 64 GiB of off-heap memory, which gives each task 4 GiB
of native memory. Both add 10 GiB of memory overhead.
+
+Both engines share these settings: G1 GC, Kryo,
`spark.io.compression.codec=zstd`, `spark.memory.fraction=0.6`,
`spark.memory.storageFraction=0.2`, `spark.sql.sources.ignoreDataLocality=true`
and `spark.executor.memoryOverhead=10g`. Comet adds:
+
+```
+spark.executor.memory=32g
+spark.memory.offHeap.enabled=true
+spark.memory.offHeap.size=64g
+spark.plugins=org.apache.spark.CometPlugin
+spark.shuffle.manager=org.apache.spark.sql.comet.execution.shuffle.CometShuffleManager
+spark.comet.parquet.prefetchNextFileMetadata.enabled=true
+spark.comet.scan.icebergNative.enabled=true
+spark.comet.scan.icebergNative.dataFileConcurrencyLimit=4
+spark.executor.defaultJavaOptions=-XX:TrimNativeHeapInterval=5000
-Xlog:trimnative
+```
+
+Two of these settings were needed to run the full suite on Comet:
+
+- **Off-heap memory.** With a 64 GiB heap and 32 GiB of off-heap memory, each
task gets 2 GiB of native memory. That is not enough for q21's sort-merge join
stage on Comet 1.2.0, which allocates about 2.2 GiB per task and reserves up to
3.1 GiB, so tasks failed and were retried.
+- **`-XX:TrimNativeHeapInterval`.** Without it, glibc keeps native memory that
Comet has freed, and executor resident memory grows from one query to the next.
Late in the suite, q21 had 11 executors OOMKilled while Comet's memory pools
were well inside their limits. Trimming every 5 s kept executors at 61 to 70
GiB of resident memory and cost about 4% on the other queries. The Comet tuning
guide now recommends it
([#7342](https://github.com/apache/datafusion-comet/pull/7342)).
+
+## What nested benchmarks revealed
+
+The nested schema uses single-element arrays of structs, with no maps for now.
Even that was enough to expose issues that were invisible on flat TPC-H. In our
first runs, Comet was slower than Spark on q20 and q21, and barely faster on
q02. Tracing those queries surfaced a lot of technical debt in how nested data
is read, shuffled and spilled, in Comet and in the iceberg-rust reader
underneath it.
+
+The issues fell into four groups:
+
+- **Spilling.** Wide nested values made q21's sort spill several hundred GiB
per iteration, and Comet wrote spill files uncompressed.
+- **Reading only what a query needs.** Comet's native Iceberg scan decoded
every field of a nested column, even when Spark had pruned it to one field.
This is why q20 was no faster than Spark in our first runs.
+- **Shuffle.** Round-robin shuffle wrote wide nested fields two to three times
slower than Spark. The remaining nested shuffle work is still open.
+- **Nested reads in iceberg-rust.** Nested fields read as NULL, some list
types could not be read at all, and projection cost grew with the number of
nested leaves.
+
+Spills and shuffle writes happen in flat workloads too. The nested schema
changed the data volume or access pattern enough to expose them.
+
+The issues and pull requests, with their measurements and details, are listed
in [Issues and pull requests](#issues-and-pull-requests) at the end of the post.
+
+## Results
+
+### Total execution time
+
+Comet 1.1.0 takes 869 s against 1,519 s for Spark (1.75x) and is slower than
Spark on one query, q14 (8.8 s against 8.1 s). Comet 1.2.0 takes 790 s (1.92x),
9% less than 1.1.0, and is faster than Spark on all 22 queries.
+
+### Execution time by query
+
+<img
+src="/blog/images/comet-nested-tpch/results-queries.svg"
+width="100%"
+class="img-fluid"
+alt="Execution time by query for Spark 4.1.3, Comet 1.1.0 and upcoming Comet
1.2.0"
+/>
+
+| Query | Spark 4.1.3 (s) | Comet 1.1.0 (s) | Comet 1.2.0, upcoming (s) |
Speedup, 1.1.0 | Speedup, 1.2.0 |
+| ----- | ---: | ---: | ---: | ---: | ---: |
+| q01 | 84.9 | 17.0 | 14.8 | 5.01x | 5.73x |
+| q02 | 39.3 | 23.4 | 27.0 | 1.68x | 1.45x |
+| q03 | 39.7 | 21.6 | 20.1 | 1.83x | 1.97x |
+| q04 | 45.2 | 19.8 | 16.1 | 2.28x | 2.81x |
+| q05 | 82.6 | 48.5 | 43.9 | 1.70x | 1.88x |
+| q06 | 5.0 | 2.2 | 2.1 | 2.22x | 2.41x |
+| q07 | 58.6 | 35.3 | 33.5 | 1.66x | 1.75x |
+| q08 | 84.6 | 64.1 | 59.3 | 1.32x | 1.43x |
+| q09 | 97.1 | 73.2 | 64.9 | 1.33x | 1.50x |
+| q10 | 50.9 | 29.2 | 26.6 | 1.74x | 1.91x |
+| q11 | 59.9 | 41.8 | 40.0 | 1.43x | 1.50x |
+| q12 | 37.1 | 17.8 | 15.2 | 2.08x | 2.44x |
+| q13 | 25.6 | 9.3 | 8.7 | 2.75x | 2.95x |
+| q14 | 8.1 | 8.8 | 4.3 | 0.93x | 1.91x |
+| q15 | 34.2 | 21.1 | 23.0 | 1.62x | 1.49x |
+| q16 | 31.9 | 14.7 | 15.4 | 2.16x | 2.08x |
+| q17 | 133.6 | 48.1 | 42.8 | 2.78x | 3.13x |
+| q18 | 152.6 | 76.3 | 56.4 | 2.00x | 2.71x |
+| q19 | 32.1 | 17.2 | 15.0 | 1.87x | 2.14x |
+| q20 | 34.3 | 24.6 | 24.3 | 1.39x | 1.41x |
+| q21 | 332.0 | 218.9 | 202.1 | 1.52x | 1.64x |
+| q22 | 49.8 | 36.2 | 35.0 | 1.38x | 1.43x |
+| **Total** | **1,519** | **869** | **790** | **1.75x** | **1.92x** |
+
+The largest gains from 1.1.0 to 1.2.0 are q14 (2.0x), q18 (1.35x) and q04
(1.23x). Comet 1.1.0 is slightly faster than 1.2.0 on q02, q15 and q16. Each
version ran once, so differences this small may be noise.
Review Comment:
This conflicts with the setup section (line 114), which says each query runs
three times and the figures are the fastest iteration. "Each version ran once"
reads as if there was a single iteration. Could we say that each version was
run in one Spark application and the figures are best-of-3 within it? That also
makes the noise caveat more precise, since the three iterations share one
application.
##########
content/blog/2026-10-25-comet-nested-tpch.md:
##########
@@ -0,0 +1,225 @@
+---
+layout: post
+title: "Nested TPC-H: A Benchmark for Nested Data in Apache DataFusion Comet"
+date: 2026-10-25
+author: pmc
+categories: [performance, subprojects]
+---
+
+<!--
+{% comment %}
+Licensed to the Apache Software Foundation (ASF) under one or more
+contributor license agreements. See the NOTICE file distributed with
+this work for additional information regarding copyright ownership.
+The ASF licenses this file to you under the Apache License, Version 2.0
+(the "License"); you may not use this file except in compliance with
+the License. You may obtain a copy of the License at
+
+http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+{% endcomment %}
+-->
+
+[TOC]
+
+Apache DataFusion Comet tracked its performance with TPC-H and TPC-DS, and the
numbers were encouraging. Real user pipelines showed smaller gains, or none at
all. What those pipelines had in common was nested data, which neither
benchmark contains.
+
+To measure that gap, we built nested TPC-H. At scale factor 1000 (1 TB), the
upcoming Comet 1.2.0 runs all 22 of its queries in 790 s, against 1,519 s for
Spark 4.1.3.
+
+<img
+src="/blog/images/comet-nested-tpch/intro-total.svg"
+width="100%"
+class="img-fluid"
+alt="Total execution time at SF1000: Spark 4.1.3 1,519 s, Comet 1.1.0 869 s,
upcoming Comet 1.2.0 790 s"
+/>
+
+Nested TPC-H is a TPC-H derived benchmark in the Comet repository
([#7044](https://github.com/apache/datafusion-comet/pull/7044)). The TPC-H data
and the questions the queries answer are unchanged. The tables are reshaped to
store most columns in arrays of structs, and the queries are rewritten to
unnest those arrays. The data and queries are engine neutral, so they can be
used to compare Spark, Comet and other engines.
+
+The first runs showed where the gaps were: some queries ran slower on Comet
than on Spark, even though Comet accelerates the same queries on flat TPC-H.
+
+## Why: TPC-H is flat, real data is not
+
+[TPC-H](https://www.tpc.org/tpch/) is one of the most cited analytical
benchmarks, and its schema is flat: eight normalized tables with primitive
columns. Its 22 queries exercise well-understood choke points such as
aggregation, join ordering and correlated subqueries (Boncz, Neumann and
Erling, ["TPC-H Analyzed"](https://doi.org/10.1007/978-3-319-04936-6_5), TPCTC
2013).
+
+Real data is often nested. Event logs, JSON, Avro and Protobuf records, and
many data lake tables are full of structs and arrays, which Parquet stores with
the record shredding scheme from Dremel (Melnik et al.,
["Dremel"](http://www.vldb.org/pvldb/vldb2010/papers/R29.pdf), VLDB 2010).
Nested data adds work that flat benchmarks never exercise:
+
+- **Reading.** Rebuilding list offsets and validity bitmaps from repetition
and definition levels, and decoding only the fields a query uses.
+- **Planning.** Nested column pruning and predicate pushdown through `explode`.
+- **Operators.** `explode`, field access, and sort, join and exchange over
wide variable-size values.
+- **Memory and spilling.** Variable-size values make memory use and spill
volume harder to predict.
+- **Conversion.** Moving nested values between a JVM and a native
representation.
+
+Comet hit the shuffle case in
[#5397](https://github.com/apache/datafusion-comet/issues/5397), where
round-robin shuffle writes of wide nested fields were two to three times slower
than Spark's. The issue notes that it was "invisible for TPC-* benchmarks
because of historical flat schema".
+
+We needed a benchmark that filled the gap between academic benchmarking and
real user workloads.
+
+## How: an adapter that nests TPC-H
+
+We did not want to invent a new benchmark from scratch. TPC-H is well
understood, widely published and easy to generate, and its business shape
(customers, orders, line items, suppliers and parts) is what makes its results
meaningful. The goal was to preserve that shape and those questions while
making every query process nested data.
+
+The idea turned out to be surprisingly simple: a transformation of both the
data and the queries that leaves the key and partition columns at the top level
and nests everything else. The partition columns are `p_brand`, `c_mktsegment`,
`o_orderdate` and `l_shipdate`, and every other non-key column moves into one
`array<struct>` named `<table>_data`. Each array holds exactly one element, so
exploding it never changes the row count and the queries return the same
answers as TPC-H. For `lineitem`:
+
+```
+-- flat -- nested
+l_orderkey BIGINT l_orderkey BIGINT
+l_partkey BIGINT l_partkey BIGINT
+l_suppkey BIGINT l_suppkey BIGINT
+l_linenumber INT l_shipdate DATE -- partition
column
+l_quantity DECIMAL(12,2) lineitem_data ARRAY<STRUCT<
+l_extendedprice DECIMAL(12,2) l_linenumber: INT,
+l_discount DECIMAL(12,2) l_quantity: DECIMAL(12,2),
+l_tax DECIMAL(12,2) l_extendedprice: DECIMAL(12,2),
+l_returnflag STRING l_discount: DECIMAL(12,2),
+l_linestatus STRING l_tax: DECIMAL(12,2),
+l_shipdate DATE l_returnflag: STRING,
+l_commitdate DATE l_linestatus: STRING,
+l_receiptdate DATE l_commitdate: DATE,
+l_shipinstruct STRING l_receiptdate: DATE,
+l_shipmode STRING l_shipinstruct: STRING,
+l_comment STRING l_shipmode: STRING,
+ l_comment: STRING>>
+```
+
+Each of the 22 queries is rewritten to unnest only the arrays it needs. The
queries are in
[`benchmarks/nested/queries`](https://github.com/apache/datafusion-comet/tree/main/benchmarks/nested/queries).
+
+```sql
+from (select l_orderkey, l_partkey, l_suppkey, l_shipdate,
+ explode(lineitem_data) as lineitem_data from lineitem) as lineitem
+where lineitem_data.l_quantity < 24
+```
+
+[`nest_tpch.scala`](https://github.com/apache/datafusion-comet/blob/main/benchmarks/nested/scripts/nest_tpch.scala)
converts an existing flat TPC-H dataset to the nested layout. The source and
the target can each be Parquet or Iceberg. The script runs under `spark-shell`
on a laptop or a cluster, casts columns to fixed types, writes zstd-compressed
Parquet with 512 MiB row groups, partitions by the columns above and checks row
counts:
+
+```bash
+spark-shell -i scripts/nest_tpch.scala \
+ --conf spark.nesttpch.source=parquet:/data/tpch/sf10 \
+ --conf spark.nesttpch.target=parquet:/data/tpch-nested/sf10
+```
+
+The
[README](https://github.com/apache/datafusion-comet/tree/main/benchmarks/nested)
has the Iceberg variants and all settings.
+
+The script and queries in the Comet repository also serve as an example. You
can customize them to test any nested complexity, such as multi-element arrays,
maps or deeper nesting.
+
+## Benchmark setup
+
+- **Data**: nested TPC-H at scale factor 1000 (1 TB), created from flat TPC-H
with `nest_tpch.scala`.
+- **Tables**: Apache Iceberg tables in a Hadoop catalog on S3, with
zstd-compressed Parquet data files, 512 MiB row groups and the partitioning
above. Comet reads them with its native Iceberg scan, which is built on
[iceberg-rust](https://github.com/apache/iceberg-rust).
+- **Versions**: Spark 4.1.3 (Scala 2.13, JDK 17) and Iceberg 1.11.0. Comet
1.1.0, and the upcoming Comet 1.2.0 built from `branch-1.2` at `746246049`. The
`1.2.0-rc1` tag is five commits ahead of that build, all backported bug fixes
that we don't expect to change these numbers. Both Comet builds use the native
Iceberg scan and the default allocator (glibc).
+- **Cluster**: Kubernetes, with 32 executors of 16 task slots each (8 CPUs per
executor pod), shuffle and spill files on a block volume attached to each
executor, and `spark.sql.shuffle.partitions=512`.
+- **Method**: the queries in `benchmarks/nested`, run from `spark-shell`. Each
query runs three times in one application with no warm-up, timed from
`spark.sql()` to the end of `collect()`. Figures are the fastest iteration, and
totals are sums of the per-query times. Every query returned the expected TPC-H
row count.
+- **Memory**: Spark runs with a 64 GiB heap and no off-heap memory. Comet runs
with a 32 GiB heap and 64 GiB of off-heap memory, which gives each task 4 GiB
of native memory. Both add 10 GiB of memory overhead.
+
+Both engines share these settings: G1 GC, Kryo,
`spark.io.compression.codec=zstd`, `spark.memory.fraction=0.6`,
`spark.memory.storageFraction=0.2`, `spark.sql.sources.ignoreDataLocality=true`
and `spark.executor.memoryOverhead=10g`. Comet adds:
+
+```
+spark.executor.memory=32g
+spark.memory.offHeap.enabled=true
+spark.memory.offHeap.size=64g
+spark.plugins=org.apache.spark.CometPlugin
+spark.shuffle.manager=org.apache.spark.sql.comet.execution.shuffle.CometShuffleManager
+spark.comet.parquet.prefetchNextFileMetadata.enabled=true
+spark.comet.scan.icebergNative.enabled=true
+spark.comet.scan.icebergNative.dataFileConcurrencyLimit=4
+spark.executor.defaultJavaOptions=-XX:TrimNativeHeapInterval=5000
-Xlog:trimnative
+```
+
+Two of these settings were needed to run the full suite on Comet:
+
+- **Off-heap memory.** With a 64 GiB heap and 32 GiB of off-heap memory, each
task gets 2 GiB of native memory. That is not enough for q21's sort-merge join
stage on Comet 1.2.0, which allocates about 2.2 GiB per task and reserves up to
3.1 GiB, so tasks failed and were retried.
+- **`-XX:TrimNativeHeapInterval`.** Without it, glibc keeps native memory that
Comet has freed, and executor resident memory grows from one query to the next.
Late in the suite, q21 had 11 executors OOMKilled while Comet's memory pools
were well inside their limits. Trimming every 5 s kept executors at 61 to 70
GiB of resident memory and cost about 4% on the other queries. The Comet tuning
guide now recommends it
([#7342](https://github.com/apache/datafusion-comet/pull/7342)).
+
+## What nested benchmarks revealed
+
+The nested schema uses single-element arrays of structs, with no maps for now.
Even that was enough to expose issues that were invisible on flat TPC-H. In our
first runs, Comet was slower than Spark on q20 and q21, and barely faster on
q02. Tracing those queries surfaced a lot of technical debt in how nested data
is read, shuffled and spilled, in Comet and in the iceberg-rust reader
underneath it.
+
+The issues fell into four groups:
+
+- **Spilling.** Wide nested values made q21's sort spill several hundred GiB
per iteration, and Comet wrote spill files uncompressed.
+- **Reading only what a query needs.** Comet's native Iceberg scan decoded
every field of a nested column, even when Spark had pruned it to one field.
This is why q20 was no faster than Spark in our first runs.
+- **Shuffle.** Round-robin shuffle wrote wide nested fields two to three times
slower than Spark. The remaining nested shuffle work is still open.
+- **Nested reads in iceberg-rust.** Nested fields read as NULL, some list
types could not be read at all, and projection cost grew with the number of
nested leaves.
+
+Spills and shuffle writes happen in flat workloads too. The nested schema
changed the data volume or access pattern enough to expose them.
+
+The issues and pull requests, with their measurements and details, are listed
in [Issues and pull requests](#issues-and-pull-requests) at the end of the post.
+
+## Results
+
+### Total execution time
+
+Comet 1.1.0 takes 869 s against 1,519 s for Spark (1.75x) and is slower than
Spark on one query, q14 (8.8 s against 8.1 s). Comet 1.2.0 takes 790 s (1.92x),
9% less than 1.1.0, and is faster than Spark on all 22 queries.
+
+### Execution time by query
+
+<img
+src="/blog/images/comet-nested-tpch/results-queries.svg"
+width="100%"
+class="img-fluid"
+alt="Execution time by query for Spark 4.1.3, Comet 1.1.0 and upcoming Comet
1.2.0"
+/>
+
+| Query | Spark 4.1.3 (s) | Comet 1.1.0 (s) | Comet 1.2.0, upcoming (s) |
Speedup, 1.1.0 | Speedup, 1.2.0 |
+| ----- | ---: | ---: | ---: | ---: | ---: |
+| q01 | 84.9 | 17.0 | 14.8 | 5.01x | 5.73x |
+| q02 | 39.3 | 23.4 | 27.0 | 1.68x | 1.45x |
+| q03 | 39.7 | 21.6 | 20.1 | 1.83x | 1.97x |
+| q04 | 45.2 | 19.8 | 16.1 | 2.28x | 2.81x |
+| q05 | 82.6 | 48.5 | 43.9 | 1.70x | 1.88x |
+| q06 | 5.0 | 2.2 | 2.1 | 2.22x | 2.41x |
+| q07 | 58.6 | 35.3 | 33.5 | 1.66x | 1.75x |
+| q08 | 84.6 | 64.1 | 59.3 | 1.32x | 1.43x |
+| q09 | 97.1 | 73.2 | 64.9 | 1.33x | 1.50x |
+| q10 | 50.9 | 29.2 | 26.6 | 1.74x | 1.91x |
+| q11 | 59.9 | 41.8 | 40.0 | 1.43x | 1.50x |
+| q12 | 37.1 | 17.8 | 15.2 | 2.08x | 2.44x |
+| q13 | 25.6 | 9.3 | 8.7 | 2.75x | 2.95x |
+| q14 | 8.1 | 8.8 | 4.3 | 0.93x | 1.91x |
+| q15 | 34.2 | 21.1 | 23.0 | 1.62x | 1.49x |
+| q16 | 31.9 | 14.7 | 15.4 | 2.16x | 2.08x |
+| q17 | 133.6 | 48.1 | 42.8 | 2.78x | 3.13x |
+| q18 | 152.6 | 76.3 | 56.4 | 2.00x | 2.71x |
+| q19 | 32.1 | 17.2 | 15.0 | 1.87x | 2.14x |
+| q20 | 34.3 | 24.6 | 24.3 | 1.39x | 1.41x |
+| q21 | 332.0 | 218.9 | 202.1 | 1.52x | 1.64x |
+| q22 | 49.8 | 36.2 | 35.0 | 1.38x | 1.43x |
+| **Total** | **1,519** | **869** | **790** | **1.75x** | **1.92x** |
+
+The largest gains from 1.1.0 to 1.2.0 are q14 (2.0x), q18 (1.35x) and q04
(1.23x). Comet 1.1.0 is slightly faster than 1.2.0 on q02, q15 and q16. Each
version ran once, so differences this small may be noise.
Review Comment:
q02 goes from 23.4 s on 1.1.0 to 27.0 s on 1.2.0 (1.68x to 1.45x), about 15%
slower, while q15 and q16 differ by 9% and 5%. Grouping all three as "slightly
faster" on 1.1.0 undersells q02. Either call it out as a regression, or if we
think it is noise, say so with some support (e.g. spread across the three
iterations).
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
