Re: VLDB RTO issue

From: Mladen Gogala <gogala.mladen_at_gmail.com>
Date: Sat, 22 Feb 2020 10:22:55 -0500
Message-ID: <15700b33-f358-6b2e-6d63-0e177f80a4ca_at_gmail.com>


You can do that with specialized hardware, like all flash arrays and fast Mellanox 100Gb Ethernet. From my experience, the worst bottleneck is the network, followed by the disk speeds. I have been routinely achieving 3.5 TB/hr on a 10Gb network. There was even a client that needed an even faster restore, so they used 2 10 Gb bonded interfaces. The thing scaled linearly and they got 7 TB per hour. That would make  RMAN restore even possible. But my favorite method, by far, are disk snapshot technologies, like SnapVault or SRDF.

On 2/21/20 3:07 PM, Mark J. Bobak wrote:
> If you're strictly relying on RMAN (or traditional non-RMAN) backups,
> copying files, I'm not sure you can restore 13 TB in two hours.  I
> guess there's a chance, if you have the CPU and I/O resources and run
> recovery in parallel.

-- 
Mladen Gogala
Database Consultant
Tel: (347) 321-1217

--
http://www.freelists.org/webpage/oracle-l
Received on Sat Feb 22 2020 - 16:22:55 CET

Original text of this message