[
https://issues.apache.org/jira/browse/HIVE-15147?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Sergey Shelukhin updated HIVE-15147:
------------------------------------
Description:
The primary goal for the first pass is caching text formats. Nothing would
prevent other formats from using the same path, in principle, although, as was
originally done with ORC, it may be better to have native caching support
optimized for each particular format.
Given that caching pure text is not smart, and we already have ORC-encoded
cache that is columnar due to ORC file structure, we will try to reuse that.
The general idea is to treat all the data in the world as merely ORC that was
compressed with some poor compression codec, such as csv. Using the original IF
and serde, as well as ORC writer (with some heavyweight optimizations removed,
potentially), we can "uncompress" the data into "original" ORC, then reuse a
lot of the existing code.
Various other points:
1) Granularity in the file will have to be somehow determined (horizontal
slicing of the file, to avoid caching entire columns). We can base it on
arbitrary disk offsets determined during reading, but they will actually have
to be propagated to the reader from the original inputformat. Row counts are
easier to use but there's a problem of how to actually map them to missing
ranges to read from disk.
2) Obviously for row-based formats, if any one column one needs is evicted,
"all the columns" have to be read for the corresponding slice. The vague plan
is to handle this implicitly, similarly to how ORC reader handles CB-RG
overlaps - it will just so happen that a missing column will expand the
disk-range-to-read into the whole horizontal slice of the file.
3) Granularity/etc. won't work for gzipped text. If anything at all is evicted,
the entire file has to be re-read. Gzipped text is a ridiculous feature, so
this is by design.
4) In future, it would be possible to also build some form or metadata/indexes
for this cached data to do PPD, etc. This is out of the scope of this stage.
was:
The primary target for the first pass is caching text formats. Nothing would
prevent other formats from using the same path, in principle, although, as was
originally done with ORC, it may be better to have native caching support
optimized for each particular format.
Given that caching pure text is not smart, and we already have ORC-encoded
cache that is columnar due to ORC file structure, we will try to reuse that.
The general idea is to treat all the data in the world as merely ORC that was
compressed with some poor compression codec, such as csv. Using the original IF
and serde, as well as ORC writer (with some heavyweight optimizations removed,
potentially), we can "uncompress" the data into "original" ORC, then reuse a
lot of the existing code.
Various other points:
1) Granularity in the file will have to be somehow determined (horizontal
slicing of the file, to avoid caching entire columns). We can base it on
arbitrary disk offsets determined during reading, but they will actually have
to be propagated to the reader from the original inputformat. Row counts are
easier to use but there's a problem of how to actually map them to missing
ranges to read from disk.
2) Obviously for row-based formats, if any one column one needs is evicted,
"all the columns" have to be read for the corresponding slice. The vague plan
is to handle this implicitly, similarly to how ORC reader handles CB-RG
overlaps - it will just so happen that a missing column will expand the
disk-range-to-read into the whole horizontal slice of the file.
3) Granularity/etc. won't work for gzipped text. If anything at all is evicted,
the entire file has to be re-read. Gzipped text is a ridiculous feature, so
this is by design.
4) In future, it would be possible to also build some form or metadata/indexes
for this cached data to do PPD, etc. This is out of the scope of this stage.
> LLAP: use LLAP cache for non-columnar formats in a somewhat general way
> -----------------------------------------------------------------------
>
> Key: HIVE-15147
> URL: https://issues.apache.org/jira/browse/HIVE-15147
> Project: Hive
> Issue Type: Bug
> Reporter: Sergey Shelukhin
> Assignee: Sergey Shelukhin
>
> The primary goal for the first pass is caching text formats. Nothing would
> prevent other formats from using the same path, in principle, although, as
> was originally done with ORC, it may be better to have native caching support
> optimized for each particular format.
> Given that caching pure text is not smart, and we already have ORC-encoded
> cache that is columnar due to ORC file structure, we will try to reuse that.
> The general idea is to treat all the data in the world as merely ORC that was
> compressed with some poor compression codec, such as csv. Using the original
> IF and serde, as well as ORC writer (with some heavyweight optimizations
> removed, potentially), we can "uncompress" the data into "original" ORC, then
> reuse a lot of the existing code.
> Various other points:
> 1) Granularity in the file will have to be somehow determined (horizontal
> slicing of the file, to avoid caching entire columns). We can base it on
> arbitrary disk offsets determined during reading, but they will actually have
> to be propagated to the reader from the original inputformat. Row counts are
> easier to use but there's a problem of how to actually map them to missing
> ranges to read from disk.
> 2) Obviously for row-based formats, if any one column one needs is evicted,
> "all the columns" have to be read for the corresponding slice. The vague plan
> is to handle this implicitly, similarly to how ORC reader handles CB-RG
> overlaps - it will just so happen that a missing column will expand the
> disk-range-to-read into the whole horizontal slice of the file.
> 3) Granularity/etc. won't work for gzipped text. If anything at all is
> evicted, the entire file has to be re-read. Gzipped text is a ridiculous
> feature, so this is by design.
> 4) In future, it would be possible to also build some form or
> metadata/indexes for this cached data to do PPD, etc. This is out of the
> scope of this stage.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
