Comments on: I/O Operations, Arguably the Only Performance Metric You Need http://michaeljswart.com/?p=587 Database Whisperer Mon, 06 Sep 2010 00:56:15 +0000 hourly 1 http://wordpress.org/?v=3.0.1 By: Something for the Weekend: SQL Server Links 12/03/10 | John Sansom - SQL Server DBA in the UK http://michaeljswart.com/?p=587&cpage=1#comment-2879 Something for the Weekend: SQL Server Links 12/03/10 | John Sansom - SQL Server DBA in the UK Sat, 17 Apr 2010 08:22:12 +0000 http://michaeljswart.com/?p=587#comment-2879 [...] I/O Operations, Arguably the Only Performance Metric You Need – It’s that Database Whisperer chap again, Michael J Swart discusses. [...] [...] I/O Operations, Arguably the Only Performance Metric You Need – It’s that Database Whisperer chap again, Michael J Swart discusses. [...]

]]> By: Michael J. Swart http://michaeljswart.com/?p=587&cpage=1#comment-2493 Michael J. Swart Thu, 25 Mar 2010 16:20:46 +0000 http://michaeljswart.com/?p=587#comment-2493 Good points Richard, and I think in that case striking the balance is the right approach. In general everyone should be aware of absolute rules of thumb. I deliberately try to write blog articles in black and white rather than shades of gray in the hopes that someone calls me on a particular point. By defending or (yikes) conceding points, I learn a topic deeper. Good points Richard, and I think in that case striking the balance is the right approach.

In general everyone should be aware of absolute rules of thumb. I deliberately try to write blog articles in black and white rather than shades of gray in the hopes that someone calls me on a particular point. By defending or (yikes) conceding points, I learn a topic deeper.

]]> By: Richard http://michaeljswart.com/?p=587&cpage=1#comment-2490 Richard Thu, 25 Mar 2010 16:08:21 +0000 http://michaeljswart.com/?p=587#comment-2490 Tricky point ( about laying off the CPU heavy stuff ) I think that striking the balance between retrieving enough information and presenting it to the developer for manipulation, and what the database can retrieve and do the CPU work on... is the tricky part for me. Probably my two biggest defenses for this are... 1) lists ( ie. paging ) 2) retrieving details for x expected entries ( ie. app wants id IN (1, 102, 201, 450001 ) ) Both of which require some amount of heavy lifting by the server. ( at least in my implementations ) Tricky point ( about laying off the CPU heavy stuff )

I think that striking the balance between retrieving enough information and presenting it to the developer for manipulation, and what the database can retrieve and do the CPU work on… is the tricky part for me. Probably my two biggest defenses for this are…

1) lists ( ie. paging )

2) retrieving details for x expected entries ( ie. app wants id IN (1, 102, 201, 450001 ) )

Both of which require some amount of heavy lifting by the server. ( at least in my implementations )

]]> By: T-SQL Tuesday #4 - IO, IO It's Off To Disk We Go | SQL Server Blog - StraightPath Solutions http://michaeljswart.com/?p=587&cpage=1#comment-2309 T-SQL Tuesday #4 - IO, IO It's Off To Disk We Go | SQL Server Blog - StraightPath Solutions Thu, 11 Mar 2010 04:41:10 +0000 http://michaeljswart.com/?p=587#comment-2309 [...] Michael Swart goes to the other side of the coin a bit. He talks about his own experiences and that tuning to IO is one of the best ways he has seen to improve query performance. I think if you take his advice, cross apply some of Kalen’s and Rob’s advice you’ll do just fine. I agree with Michael, in the majority of cases I come across, reducing the reads reduces the duration. Just be cautious like Kalen said and don’t go overboard like Rob said. [...] [...] Michael Swart goes to the other side of the coin a bit. He talks about his own experiences and that tuning to IO is one of the best ways he has seen to improve query performance. I think if you take his advice, cross apply some of Kalen’s and Rob’s advice you’ll do just fine. I agree with Michael, in the majority of cases I come across, reducing the reads reduces the duration. Just be cautious like Kalen said and don’t go overboard like Rob said. [...]

]]> By: Jon Russell http://michaeljswart.com/?p=587&cpage=1#comment-2292 Jon Russell Tue, 09 Mar 2010 21:31:51 +0000 http://michaeljswart.com/?p=587#comment-2292 Nice post Michael. Nice post Michael.

]]> By: Dave Schutz http://michaeljswart.com/?p=587&cpage=1#comment-2288 Dave Schutz Tue, 09 Mar 2010 16:59:44 +0000 http://michaeljswart.com/?p=587#comment-2288 Michael, Nice thought looking from the hardware view of things. It will be interesting to see if this also applies to SSD. Michael,

Nice thought looking from the hardware view of things. It will be interesting to see if this also applies to SSD.

]]> By: Michael J. Swart http://michaeljswart.com/?p=587&cpage=1#comment-2286 Michael J. Swart Tue, 09 Mar 2010 15:59:18 +0000 http://michaeljswart.com/?p=587#comment-2286 Thanks, I'll fix that soon. (replace > with >) Thanks, I’ll fix that soon. (replace > with >)

]]> By: Aaron Bertrand http://michaeljswart.com/?p=587&cpage=1#comment-2284 Aaron Bertrand Tue, 09 Mar 2010 15:04:28 +0000 http://michaeljswart.com/?p=587#comment-2284 Nice post. Your sample query's HTML deceives us though: it replaced > with > (second-last line). Cheers, Aaron Nice post. Your sample query’s HTML deceives us though: it replaced > with > (second-last line).

Cheers,
Aaron

]]> By: Stefan http://michaeljswart.com/?p=587&cpage=1#comment-2283 Stefan Tue, 09 Mar 2010 13:31:55 +0000 http://michaeljswart.com/?p=587#comment-2283 Good post, glad I stumbled upon the #tsql2sday lots of good stuff today. Thanks! Good post, glad I stumbled upon the #tsql2sday lots of good stuff today.
Thanks!

]]> By: Michael J. Swart http://michaeljswart.com/?p=587&cpage=1#comment-2282 Michael J. Swart Tue, 09 Mar 2010 12:39:08 +0000 http://michaeljswart.com/?p=587#comment-2282 Thanks Pinal, That's especially flattering coming from you. Thanks Pinal, That’s especially flattering coming from you.

]]>