diff --git a/src/common/recordbatch/src/error.rs b/src/common/recordbatch/src/error.rs
index 49bfe02bf1d2..b2c88fffee9a 100644
--- a/src/common/recordbatch/src/error.rs
+++ b/src/common/recordbatch/src/error.rs
@@ -37,7 +37,7 @@ pub enum Error {
         source: datatypes::error::Error,
     },
 
-    #[snafu(display("External error, source: {}", source))]
+    #[snafu(display("External error, location: {}, source: {}", location, source))]
     External {
         location: Location,
         source: BoxedError,
diff --git a/src/mito2/src/read.rs b/src/mito2/src/read.rs
index 0c917408471f..9c6f4593be8a 100644
--- a/src/mito2/src/read.rs
+++ b/src/mito2/src/read.rs
@@ -16,6 +16,7 @@
 
 pub(crate) mod scan_region;
 pub mod seq_scan;
+pub(crate) mod stream;
 
 use std::sync::Arc;
 
diff --git a/src/mito2/src/read/scan_region.rs b/src/mito2/src/read/scan_region.rs
index 1bdf30fc14b6..1d6095e2777f 100644
--- a/src/mito2/src/read/scan_region.rs
+++ b/src/mito2/src/read/scan_region.rs
@@ -14,9 +14,7 @@
 
 //! Scans a region according to the scan request.
 
-use std::collections::HashMap;
-
-use common_recordbatch::SendableRecordBatchStream;
+use common_recordbatch::{SendableRecordBatchStream};
 use common_telemetry::debug;
 use common_time::range::TimestampRange;
 use object_store::ObjectStore;
diff --git a/src/mito2/src/read/seq_scan.rs b/src/mito2/src/read/seq_scan.rs
index b1737f89a661..c2bea72080d3 100644
--- a/src/mito2/src/read/seq_scan.rs
+++ b/src/mito2/src/read/seq_scan.rs
@@ -49,6 +49,7 @@ pub struct SeqScan {
 
 impl SeqScan {
     /// Creates a new [SeqScan].
+    #[must_use]
     pub fn new(metadata: RegionMetadataRef, file_dir: &str, object_store: ObjectStore) -> SeqScan {
         SeqScan {
             metadata,
@@ -63,39 +64,47 @@ impl SeqScan {
     }
 
     /// Set projection.
+    #[must_use]
     pub fn with_projection(mut self, projection: Option<Vec<ColumnId>>) -> Self {
         self.projection = projection;
         self
     }
 
     /// Set time range filter for time index.
+    #[must_use]
     pub fn with_time_range(mut self, time_range: Option<TimestampRange>) -> Self {
         self.time_range = time_range;
         self
     }
 
     /// Set predicate to push down.
+    #[must_use]
     pub fn with_predicate(mut self, predicate: Option<Predicate>) -> Self {
         self.predicate = predicate;
         self
     }
 
     /// Set memtables to read.
+    #[must_use]
     pub fn with_memtables(mut self, memtables: Vec<MemtableRef>) -> Self {
         self.memtables = memtables;
         self
     }
 
     /// Set files to read.
+    #[must_use]
     pub fn with_files(mut self, files: Vec<FileHandle>) -> Self {
         self.files = files;
         self
     }
 
     /// Builds a stream for the query.
+    #[must_use]
     pub fn build(&self) -> SendableRecordBatchStream {
         // Scans all memtables and SSTs.
         // Builds a merge reader to merge results.
+
+        //
         unimplemented!()
     }
 }
diff --git a/src/mito2/src/read/stream.rs b/src/mito2/src/read/stream.rs
new file mode 100644
index 000000000000..2a33471dbff4
--- /dev/null
+++ b/src/mito2/src/read/stream.rs
@@ -0,0 +1,187 @@
+// Copyright 2023 Greptime Team
+//
+// Licensed 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.
+
+//! Record batch stream.
+
+use std::sync::Arc;
+use std::pin::Pin;
+use std::task::Context;
+use std::task::Poll;
+
+use api::v1::SemanticType;
+use common_error::ext::BoxedError;
+use common_recordbatch::RecordBatchStream;
+use common_recordbatch::error::ExternalSnafu;
+use common_recordbatch::{RecordBatch};
+use datatypes::prelude::{ConcreteDataType, DataType};
+use datatypes::schema::{SchemaRef, Schema};
+use datatypes::value::ValueRef;
+use datatypes::vectors::VectorRef;
+use futures::Stream;
+use snafu::ResultExt;
+use store_api::metadata::RegionMetadata;
+use crate::read::Batch;
+use crate::row_converter::{McmpRowCodec, SortField, RowCodec};
+use crate::error::Result;
+
+/// Record batch stream implementation.
+pub(crate) struct StreamImpl<S> {
+    /// [Batch] stream.
+    stream: S,
+    /// Converts [Batch]es from the `stream` to [RecordBatch].
+    converter: BatchConverter,
+}
+
+impl<S> StreamImpl<S> {
+    /// Returns a new stream from a batch stream.
+    pub(crate) fn new(stream: S, converter: BatchConverter) -> StreamImpl<S> {
+        StreamImpl {
+            stream,
+            converter,
+        }
+    }
+}
+
+impl<S: Stream<Item = Result<Batch>> + Unpin> Stream for StreamImpl<S> {
+    type Item = common_recordbatch::error::Result<RecordBatch>;
+
+    fn poll_next(mut self: Pin<&mut Self>, cx: &mut Context<'_>) -> Poll<Option<Self::Item>> {
+        match Pin::new(&mut self.stream).poll_next(cx) {
+            Poll::Ready(Some(res)) => {
+                let record_batch = res.map_err(BoxedError::new).context(ExternalSnafu).and_then(|batch| {
+                    self.converter.convert(&batch)
+                });
+                Poll::Ready(Some(record_batch))
+            },
+            Poll::Ready(None) => Poll::Ready(None),
+            Poll::Pending => Poll::Pending,
+        }
+    }
+}
+
+impl<S: Stream<Item = Result<Batch>> + Unpin> RecordBatchStream for StreamImpl<S> {
+    fn schema(&self) -> SchemaRef {
+        self.converter.output_schema.clone()
+    }
+}
+
+/// Converts a [Batch] to a [RecordBatch].
+pub(crate) struct BatchConverter {
+    /// Maps column in [RecordBatch] to index in [Batch].
+    batch_indices: Vec<BatchIndex>,
+    /// Decoder for primary key.
+    codec: McmpRowCodec,
+    /// Schema for converted [RecordBatch].
+    output_schema: SchemaRef,
+}
+
+impl BatchConverter {
+    /// Returns a new converter with projection.
+    ///
+    /// # Panics
+    /// Panics if any index in `projection` is out of bound.
+    pub(crate) fn new(metadata: &RegionMetadata, projection: impl Iterator<Item = usize>) -> BatchConverter {
+        let mut batch_indices = Vec::with_capacity(projection.size_hint().0);
+        let mut column_schemas = Vec::with_capacity(projection.size_hint().0);
+        for idx in projection {
+            // For each projection index, we get the column id for projection.
+            let column = &metadata.column_metadatas[idx];
+
+            // Get column index in a batch by its semantic type and column id.
+            let batch_index = match column.semantic_type {
+                SemanticType::Tag => {
+                    // Safety: It is a primary key column.
+                    let index = metadata.primary_key_index(column.column_id).unwrap();
+                    BatchIndex::Tag(index)
+                },
+                SemanticType::Timestamp => BatchIndex::Timestamp,
+                SemanticType::Field => {
+                    // Safety: It is a field column.
+                    let index = metadata.field_index(column.column_id).unwrap();
+                    BatchIndex::Field(index)
+                }
+            };
+            batch_indices.push(batch_index);
+
+            column_schemas.push(metadata.schema.column_schemas()[idx].clone());
+        }
+
+        let codec = McmpRowCodec::new(
+            metadata
+                .primary_key_columns()
+                .map(|c| SortField::new(c.column_schema.data_type.clone()))
+                .collect(),
+        );
+        // Safety: Columns come from existing schema.
+        let output_schema = Arc::new(Schema::new(column_schemas));
+
+        BatchConverter {
+            batch_indices,
+            codec,
+            output_schema,
+        }
+    }
+
+    /// Returns a new converter without projection.
+    pub(crate) fn all(metadata: &RegionMetadata) -> BatchConverter {
+        BatchConverter::new(metadata, 0..metadata.column_metadatas.len())
+    }
+
+    /// Converts a [Batch] to a [RecordBatch].
+    ///
+    /// The batch must match the `projection` using to build the converter.
+    pub(crate) fn convert(&self, batch: &Batch) -> common_recordbatch::error::Result<RecordBatch> {
+        let pk_values = self.codec.decode(batch.primary_key()).map_err(BoxedError::new).context(ExternalSnafu)?;
+
+        let mut columns = Vec::with_capacity(self.output_schema.num_columns());
+        let num_rows = batch.num_rows();
+        for (index, column_schema) in self.batch_indices.iter().zip(self.output_schema.column_schemas()) {
+            match index {
+                BatchIndex::Tag(idx) => {
+                    let value = pk_values[*idx].as_value_ref();
+                    let vector = new_repeated_vector(&column_schema.data_type, value, num_rows)?;
+                    columns.push(vector);
+                },
+                BatchIndex::Timestamp => {
+                    columns.push(batch.timestamps().clone());
+                },
+                BatchIndex::Field(idx) => {
+                    columns.push(batch.fields()[*idx].data.clone());
+                },
+            }
+        }
+
+        RecordBatch::new(self.output_schema.clone(), columns)
+    }
+}
+
+/// Index of a vector in a [Batch].
+#[derive(Debug, Clone, Copy)]
+enum BatchIndex {
+    /// Index in primary keys.
+    Tag(usize),
+    /// The time index column.
+    Timestamp,
+    /// Index in fields.
+    Field(usize),
+}
+
+/// Returns a vector with repeated values.
+fn new_repeated_vector(data_type: &ConcreteDataType, value: ValueRef, num_rows: usize) -> common_recordbatch::error::Result<VectorRef> {
+    let mut mutable_vector = data_type.create_mutable_vector(1);
+    mutable_vector.try_push_value_ref(value).map_err(BoxedError::new).context(ExternalSnafu)?;
+    // This requires an addtional allocation. TODO(yingwen): Add a way to create repeated vector to data type.
+    let base_vector = mutable_vector.to_vector();
+    Ok(base_vector.replicate(&[num_rows]))
+}
diff --git a/src/mito2/src/sst/parquet/format.rs b/src/mito2/src/sst/parquet/format.rs
index 8b17826edfe8..977604c2607f 100644
--- a/src/mito2/src/sst/parquet/format.rs
+++ b/src/mito2/src/sst/parquet/format.rs
@@ -23,6 +23,8 @@
 //! ```text
 //! field 0, field 1, ..., field N, time index, primary key, sequence, op type
 //! ```
+//!
+//! We stores fields in the same order as [RegionMetadata::field_columns()](store_api::metadata::RegionMetadata::field_columns()).
 
 use std::collections::HashMap;
 use std::sync::Arc;
diff --git a/src/store-api/src/metadata.rs b/src/store-api/src/metadata.rs
index 7c5b75e869bb..7d298557bd85 100644
--- a/src/store-api/src/metadata.rs
+++ b/src/store-api/src/metadata.rs
@@ -169,6 +169,7 @@ impl RegionMetadata {
             .map(|index| &self.column_metadatas[index])
     }
 
+    /// Returns all primary key columns.
     pub fn primary_key_columns(&self) -> impl Iterator<Item = &ColumnMetadata> {
         // safety: RegionMetadata::validate ensures every primary key exists.
         self.primary_key
@@ -183,6 +184,20 @@ impl RegionMetadata {
             .filter(|column| column.semantic_type == SemanticType::Field)
     }
 
+    /// Returns a column's index in primary key if it is a primary key column.
+    ///
+    /// This does a linear search.
+    pub fn primary_key_index(&self, column_id: ColumnId) -> Option<usize> {
+        self.primary_key.iter().position(|id| *id == column_id)
+    }
+
+    /// Returns a column's index in fields if it is a field column.
+    ///
+    /// This does a linear search.
+    pub fn field_index(&self, column_id: ColumnId) -> Option<usize> {
+        self.field_columns().position(|column| column.column_id == column_id)
+    }
+
     /// Checks whether the metadata is valid.
     fn validate(&self) -> Result<()> {
         // Id to name.