alamb commented on code in PR #340:
URL: https://github.com/apache/arrow-site/pull/340#discussion_r1158475134
##########
_posts/2023-04-02-a-journey-with-apache-arrow-part-1.md:
##########
@@ -24,7 +24,7 @@ limitations under the License.
{% endcomment %}
-->
-Apache Arrow is a technology widely adopted in big data, analytics, and
machine learning applications. This article discusses our journey with Arrow,
specifically its application to telemetry, and the challenges we encountered
while optimizing the OpenTelemetry protocol to significantly reduce bandwidth
costs. The promising results we achieved inspired us to share our insights.
This article specifically focuses on transforming data from an XYZ format into
an efficient Arrow representation that optimizes both compression ratio,
transport, and data processing. Our benchmarks thus far have shown promising
results, with compression ratio improvements ranging from 1.5x to 6x, depending
on the data type (metrics, logs, traces) and distribution. The approaches
presented for addressing these challenges may be applicable to other Arrow
domains as well. This article serves as the first installment in a two-part
series.
+Apache Arrow is a technology widely adopted in big data, analytics, and
machine learning applications. In this article, we share F5's experience with
Arrow, specifically its application to telemetry, and the challenges we
encountered while optimizing the OpenTelemetry protocol to significantly reduce
bandwidth costs. The promising results we achieved inspired us to share our
insights. This article specifically focuses on transforming relatively complex
data structure from various formats into an efficient Arrow representation that
optimizes both compression ratio, transport, and data processing. We also
explore the trade-offs between different mapping and normalization strategies,
as well as the nuances of streaming and batch communication using Arrow and
Arrow Flight. Our benchmarks thus far have shown promising results, with
compression ratio improvements ranging from 1.5x to 6x, depending on the data
type (metrics, logs, traces) and distribution. The approaches presented for addr
essing these challenges may be applicable to other Arrow domains as well. This
article serves as the first installment in a two-part series.
Review Comment:
👍
--
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]
