[
https://issues.apache.org/jira/browse/SPARK-12196?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15073573#comment-15073573
]
yucai commented on SPARK-12196:
-------------------------------
Hello Wei, nice to know you :). Some explanation about our implementation:
1. About space reserve thread, in our previous version, we do have this kind of
daemon, but we removed it finally, because based on our E2E testing, almost no
overhead in current implement. getUseableSpace is to get some meta data in file
system, most of time it is well cached by OS.
Our testing environment is 4 HSW box(72 cores, 256GB memory, 10GB Nic,
HDDs/SSDs) and running real customer case NWeight, which is to compute
associations between two vertices that are n-hop away(e.g., friend-to-friend or
video-to-video relationship for recommendation).
2. Our implementation does support shuffle data also.
> Store blocks in different speed storage devices by hierarchy way
> ----------------------------------------------------------------
>
> Key: SPARK-12196
> URL: https://issues.apache.org/jira/browse/SPARK-12196
> Project: Spark
> Issue Type: New Feature
> Components: Spark Core
> Reporter: yucai
>
> *Problem*
> Nowadays, users have both SSDs and HDDs.
> SSDs have great performance, but capacity is small. HDDs have good capacity,
> but x2-x3 lower than SSDs.
> How can we get both good?
> *Solution*
> Our idea is to build hierarchy store: use SSDs as cache and HDDs as backup
> storage.
> When Spark core allocates blocks for RDD (either shuffle or RDD cache), it
> gets blocks from SSDs first, and when SSD’s useable space is less than some
> threshold, getting blocks from HDDs.
> In our implementation, we actually go further. We support a way to build any
> level hierarchy store access all storage medias (NVM, SSD, HDD etc.).
> *Performance*
> 1. At the best case, our solution performs the same as all SSDs.
> 2. At the worst case, like all data are spilled to HDDs, no performance
> regression.
> 3. Compared with all HDDs, hierarchy store improves more than *_x1.86_* (it
> could be higher, CPU reaches bottleneck in our test environment).
> 4. Compared with Tachyon, our hierarchy store still *_x1.3_* faster. Because
> we support both RDD cache and shuffle and no extra inter process
> communication.
> *Usage*
> 1. Set the priority and threshold for each layer in
> spark.storage.hierarchyStore.
> {code}
> spark.storage.hierarchyStore='nvm 50GB,ssd 80GB'
> {code}
> It builds a 3 layers hierarchy store: the 1st is "nvm", the 2nd is "sdd", all
> the rest form the last layer.
> 2. Configure each layer's location, user just needs put the keyword like
> "nvm", "ssd", which are specified in step 1, into local dirs, like
> spark.local.dir or yarn.nodemanager.local-dirs.
> {code}
> spark.local.dir=/mnt/nvm1,/mnt/ssd1,/mnt/ssd2,/mnt/ssd3,/mnt/disk1,/mnt/disk2,/mnt/disk3,/mnt/disk4,/mnt/others
> {code}
> After then, restart your Spark application, it will allocate blocks from nvm
> first.
> When nvm's usable space is less than 50GB, it starts to allocate from ssd.
> When ssd's usable space is less than 80GB, it starts to allocate from the
> last layer.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org
For additional commands, e-mail: issues-h...@spark.apache.org
