[
https://issues.apache.org/jira/browse/AURORA-257?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13999885#comment-13999885
]
Ahmed Ali-Eldin commented on AURORA-257:
----------------------------------------
We have studied quite a few of the available open-source systems. We will
start experimenting with Suro interfaced with Kafka. Suro was chosen for a few
reasons including:
1- It is built to be elastic. An Aurora user can scale it up or down based on
the logging load.
2- It has a Kafka sink, meaning that a user can simply interface it with Kafka,
and get all the power of Kafka (interfacing with Storm, HBase, Druid,
ElasticSearch and a few others)
3- It has an S3 sink, which enables users on Amazon or who plan to use Amazon
for the analytics to do so.
4- I got a tip from someone in Netflix on its speed.
5- Both Suro and Kafka are well maintained by the open-source community.
> GSoC: Investigate distributed logging systems
> ---------------------------------------------
>
> Key: AURORA-257
> URL: https://issues.apache.org/jira/browse/AURORA-257
> Project: Aurora
> Issue Type: Task
> Components: Client
> Reporter: Mark Chu-Carroll
> Assignee: Ahmed Ali-Eldin
> Priority: Minor
> Labels: gsoc2014
>
> There are a lot of different systems that do some form of distributed systems
> logging. For aurora analytics, we'd like to select one which should be easy
> to use, require minimal configuration, and reliably gather logged information
> from multiple sources into a single, queryable repository.
> This task is to put together a list of distributed logging systems that are
> usable from both Python and Java, compatible with our apache license, and
> gather information about the tradeoffs between them. Once we have all of the
> necessary information, we should make a choice for which to use.
--
This message was sent by Atlassian JIRA
(v6.2#6252)
