[ 
https://issues.apache.org/jira/browse/HBASE-21874?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16769991#comment-16769991
 ] 

Anoop Sam John commented on HBASE-21874:
----------------------------------------

bq.So in this case, assuming device is mounted in "memory mode" described here, 
it works as an extension of main memory? But we still have to access it as a 
mmap?
Yes in that mode, the device will act as an extension of DRAM.  In this case no 
need to do any mmap. It is like the DRAM. So ByteBufferIOEngine is enough then. 
 We can configure a huge size off heap cache there.  The 
ByteBuffer.allocateDirect() calls would be allocating the chunks over the Pmem 
device rather than over DRAM.    
As said, this patch is for the other mode (App Direct).. In this mode, we have 
to create file over the device mount path.  We can always do file based 
operation also. The device is to be with DAX FS.  And the OS should be capable 
to identify the pmem device.  All the file based OS calls work normally here. 
The extra is we can do mmap and the OS will not try to bring the pages into its 
Page cache (DRAM).  Instead the CPU can directly access bytes over the device 
with its load/store instructions.   For this the pmem library APIs also not a 
mandatory item.

> Bucket cache on Persistent memory
> ---------------------------------
>
>                 Key: HBASE-21874
>                 URL: https://issues.apache.org/jira/browse/HBASE-21874
>             Project: HBase
>          Issue Type: New Feature
>          Components: BucketCache
>    Affects Versions: 3.0.0
>            Reporter: ramkrishna.s.vasudevan
>            Assignee: ramkrishna.s.vasudevan
>            Priority: Major
>             Fix For: 3.0.0
>
>         Attachments: HBASE-21874.patch, HBASE-21874.patch, 
> HBASE-21874_V2.patch, Pmem_BC.png
>
>
> Non volatile persistent memory devices are byte addressable like DRAM (for 
> eg. Intel DCPMM). Bucket cache implementation can take advantage of this new 
> memory type and can make use of the existing offheap data structures to serve 
> data directly from this memory area without having to bring the data to 
> onheap.
> The patch is a new IOEngine implementation that works with the persistent 
> memory.
> Note : Here we don't make use of the persistence nature of the device and 
> just make use of the big memory it provides.
> Performance numbers to follow. 



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

Reply via email to