Re: ODA and HCC - current state

From: Kevin Closson <>
Date: Sun, 26 Aug 2012 12:17:15 -0700 (PDT)
Message-ID: <>

> HCC is a distinct  optimization.

Actually, HCC is merely "allowed" on ZFSSA (S7000) and Pillar. There is no HCC code executing in those storage processors unlike Exadata where decompression of HCC blocks (usually) does occur down in storage.

Oracle is using a simple discovery mechanism via SNMP and if they "smell" standards-based storage (it is NFS after all) that *they* think *you* should be using then they allow the creation and access of HCC blocks.

HCC is RDBMS generic code. It will function on any NFS storage if you "accidentally" forget to apply the patche that disables it. It functions on Exadata with storage offload processing disabled and it made it all the way through Beta on customers' systems that were neither Exadata nor NFS.

I've never seen precedence for this before. Patches that disables generic RDBMS functionality based on the storage vendor from whom you purchased your standards-based storage. What's next? motherboard detection that enables or disables generic RDBMS code?  Long live Open Systems.

Some more on the matter:,8

 From: Hans Forbrich <> To:
Sent: Sunday, August 26, 2012 9:28 AM
Subject: Re: ODA and HCC - current state  

HCC is supported and indeed is a preferred option in ZFS on ODA. (Some rules apply.)  Since ZFS implements copy-on-write, HCC is a distinct optimization.

Some info at


On 26/08/2012 1:20 PM, Kamus wrote:
> I have no idea about ZFS storage, AFAIK, HCC is only available in Exadata, not in ODA.

Received on Sun Aug 26 2012 - 14:17:15 CDT

Original text of this message