I Didn’t Know That 6 – What is Wrong with this Quote?

14 12 2010

December 14, 2010

(Back to the Previous Post in the Series)

In the interest of interesting quotes, a quote that I found in the “Oracle Tuning the Definitive Reference Second Edition” book on page 541:

“Oracle speeds are very high with SSD, and SSD is also cheap at only $1k/gig USD…  Companies are now offering solid-state disk replacement for the Oracle data buffer cache to speed up I/O at the physical level…”

“Physical disk I/O is measured in milliseconds, an eternity when compared to faster operations within other server components such as network RAM and CPU speeds.  For many years, Oracle shops have been embracing solid-state disks, RAM disks that operate hundreds of time faster than old-fashioned platter technology from the 1960s.  SSD also has no channel contention, and as prices fall, SSD will eventually displace the ancient magnetic spinning platters of the last century.”

What, if anything, is wrong with the above quote? 

—————–

While my review of the book only provides an in-depth technical review of the first 200 pages of the book, this blog article series will dig into some of the pages that were not specifically included in the review. 

This blog article marks the last of the blog articles that were scheduled as a component part of the “Oracle Tuning the Definitive Reference Second Edition” book review.  Regular readers will probably recognize that I attempt to encourage people to learn from mistakes, whether your own, or those that you encounter in your day to day duties working with Oracle Database.  If the mistakes are yours, it is your choice whether you continue to make the same mistakes year after year, or if you will seek to learn from the assistance provided by others.  The last four book reviews that I posted have all had a “public opinion” portion (in separate blog articles), and I think that approach adds a new dimension to book reviews – I certainly appreciate the feedback that readers have provided.  If you are a senior DBA, or a person who enjoys digging to find the root meaning of what is stated, take a serious look at buying this book and writing a review. 

The point of blog articles like this one is not to insult authors who have spent thousands of hours carefully constructing an accurate and helpful book, but instead to suggest that readers investigate when something stated does not exactly match what one believes to be true.  It could be that the author “took a long walk down a short pier”, or that the author is revealing accurate information which simply cannot be found through other resources (and may in the process be directly contradicting information sources you have used in the past).  If you do not investigate in such cases, you may lose an important opportunity to learn something that could prove to be extremely valuable.

Other pages found during a Google search of the phrase:

  • rampant-books.com/t_oracle_ssd_disk_i_o.htm
  • praetoriate.com/t_%20oraclerac_ssd_with_rac.htm
  • dba-oracle.com/t_sun_solaris_solid_state_disk_ssd.htm
  • dba-oracle.com/t_flash_disk_drives_ssd_ram_san.htm
  • dba-oracle.com/art_dbazine_2020_p2.htm

Actions

Information

3 responses

14 12 2010
Flado

Erm…
* People are replacing the DDR RAM traditionally used for “the Oracle data buffer cache” with (Flash, I assume) SSDs? How?! And how replacing the Oracle data buffer cache with a slower storage speeds up I/O at the physical layer?
* What’s “network RAM”?
* SSD has no channel contention? How is that, when even volatile RAM has it?

14 12 2010
Charles Hooper

Flado,

Excellent points.
———

Anyone have access to some platter based hard drives from the 1960s that can do a head to head comparison with SSD drives? I know that the hard drives from the early 1990s were able to achieve transfer speeds of 1MB per second, while a Crucial C300 256GB drive that retails for roughly $550 USD is able to hit speeds ranging from 270MB per second to 355MB per second depending on the controller used in the server. (Is this perhaps a hint?)

14 12 2010
Martin Berger

I assume ” … network RAM and CPU speeds.” is missing a comma: ” … network, RAM and CPU speeds.” This might make any sense.
Unfortunately the statement is not quite specific at all.
In case of db_flash_cache_file, SSDs might be a good bridge between ram (speed&price) and hdd (speed&price). But even this must be well planned.
I read this quote like ‘if SSD is used for data files, only a little db cache is needed anymore.’ Based on that interpretation I would not compare db-cache speed to IO speed – even with a ram-disk for the data files. Once again not tested/measured I am sure the same data block will be much faster accessed in the chain “search_in_buffer -> hit” against “(search_in_buffer -> fail); tell_OS_to_do_IO; IO_is_done_by_OS; (probably_copy_block_into_buffer); -> hit”. In fact it’s even more complex, but you got the point?

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s




Follow

Get every new post delivered to your Inbox.

Join 137 other followers

%d bloggers like this: