[
https://issues.apache.org/jira/browse/JCLOUDS-1302?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16031018#comment-16031018
]
Steve Loughran commented on JCLOUDS-1302:
-----------------------------------------
bq. It would be nice if the Hadoop guys used jclouds instead of their
home-grown implementations...
interesting point. FWIW, I'd be happy with a generic jclouds client which can
bind with anything.
But bear in mind we are trying to do some fairly low-level ops for (a)
performance and (b) commit protocols, while also handling sources and dest
files which can be measured in tens, to hundreds of GB. For S3 and the Azure
stores we use the vendor SDKs, the leasing calls in Azure for atomic renames,
and consistent view over S3 coming in HADOOP-13345. The seek optimisations
force us to make decisions in the S3 client when to abort a TCP stream vs
read-and-discard to end of stream, we have different content-length header
policies optimised for column formats like ORC and Parquet (HADOOP-13203)
So: a jcloud client would be good, happy to pull in. Not going to move off our
s3a, azure wasb or adl clients tho'
> Aliyun OSS blobstore
> --------------------
>
> Key: JCLOUDS-1302
> URL: https://issues.apache.org/jira/browse/JCLOUDS-1302
> Project: jclouds
> Issue Type: New Feature
> Components: jclouds-blobstore
> Affects Versions: 2.0.1
> Reporter: zreal
>
> The OSS of the Aliyun is widely used in China. So I want to konw whether the
> jclouds can support the OSS?
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
