Meetings are like bad blocking queries – they tie up precious resources and stop other people from getting work done.  Today, let’s talk about how to optimize both of ‘em.

Minutes of typical corporate meeting

Do Your Homework Ahead of Time

If you’re going to hold a meeting, build the list of decisions or tasks you want to accomplish during the meeting.  Send out that list ahead of time before you schedule the meeting, and you might be pleasantly surprised at how much work gets done without the meeting, because:

• Some tasks really only need one person to accomplish them
• Some decisions are only made by one executive (without input)
• Some research can’t be done during a meeting
• Some people will do just about anything (including extra work) to avoid meetings

When designing your SQL Server transactions, put together an agenda.  Out of the list of tasks, try to move everything out of the transaction.  Can you run some of the queries ahead of time before BEGIN TRAN?  If so, that lessens the number of locks you need for your transaction, and makes your transaction run faster.  When examining each query in the transaction, ask yourself:

• If this data changed between the time I queried it and the time the transaction committed, would it be a problem?  If you’re fetching the customer’s bank balance before you update it, that would be bad.  If, however, you’re checking the date the customer’s account opened, it probably isn’t going to change while the transaction runs.  That might be a good candidate to move out of the transaction.
• If I had to rollback the transaction, would this query be affected?  If it’s just a select, the answer is probably not – but your mileage may vary.

This is especially important for transactions that run frequently, like hundreds of times per minute.  In one engagement, we found a common transaction to update an employee’s details.  In the beginning, it checked the employee’s supervisor’s status before updating the employee record.  By moving that query out of the transaction and doing the work ahead of time, we cut the transaction execution time by over 90% without affecting the integrity of the query.

Minimize the Number of Objects Involved

Know why most meetings and transactions are like an OK Go video?  There’s way too many things involved.

And if you think that machine is complicated, think for a second about the process of filming it.  Everything had to work perfectly, the camera had to smoothly move throughout the entire operation, the guys had to move from one position to the next to hit their mark in time, and the whole thing is choreographed in a way that you don’t see the guys moving from spot to spot.  If you like this, check out their marching band version or their famous treadmill video, and note that they’re all done in one very long take with no camera changes.  (The Rube Goldberg one does have a cut in it, though.)

The more people involved in a meeting, the tougher scheduling gets, and the more people sit around twiddling their thumbs during the meeting.  If you only need someone for 5 minutes of the 60-minute meeting, try to accomplish that goal ahead of time and free up 55 minutes of their time.

When designing your SQL transactions, be acutely aware of every table you’re inviting to the party.  SQL Server has to schedule these objects because you’re not the only one who wants to use them.  The less locks you incur, the faster your transaction can get in and get out.

Have Less Meetings (and Transactions)

Do you really need to tie up multiple resources at the exact same time in order to get your work done?

Or are you just taking the lazy, selfish approach and grabbing lots of resources just because you can?

Coworkers and tables seem like easy resources to commandeer at a moment’s notice because you don’t pay the price initially.  As your work piles up, though, performance suffers.  Making the right call ahead of time pays off in the long run.

Tuning SQL Server Transactions – and Meetings is a post from: Brent Ozar - Too Much Information.
If you'd like to quote this content on other sites, please read how to use my blog's content.

Here’s a quote from the announcement:

“At first glance, the results seem fairly clear: 81% of the 1,573 respondents want a PASS Summit on the East Coast at least every four years.”

I was expecting a strong vote to move the Summit around, but 81% – that’s amazing.  Four out of every five people agreed.  It’s hard to achieve that kind of agreement in any kind of meeting at all!  The article continues to say:

“When we look at responses from only 2008 and 2009 Summit attendees (our most successful ones by far), the number who want a future Summit outside of Seattle drops to 69%.”

Let’s think about that for a second – if we want to increase PASS Summit attendance, that means getting new people to attend.  Forget the opinions of the people who’ve already been – what about the folks who haven’t been yet?  That’s the opinions we need to hear if we want to raise attendance.  Rushabh doesn’t quote that number, but it must be higher than 81% because the average went down 12% when these people were excluded.  Wow.

Seattle. Get Used To It.

So PASS started investigating what it would take to move the Summit, and they said:

“We would not be able to achieve anywhere near the same level of support from Microsoft as we do when Summit is held in Seattle. We would lose out on at least 50% and likely 75% of Microsoft presenters, developers, and SQLCAT and CSS staff – all things a majority of survey respondents listed as important or very important.”

This is interesting, because this is where we need the raw data to answer a deeper question.  My question is, “Of people who haven’t attended the Summit before, how important is access to Microsoft staff?”  If we asked them to rank the following in order of importance, I’d be curious what they chose:

• Access to Microsoft staff
• Access to Microsoft MVPs
• Access to your peers, SQL Server community members
• Low hotel and food costs (under $150/day as opposed to over $200/day)
• Summit location nearby (within 4 hours driving distance)

When I was a DBA, I didn’t attend the PASS Summit to talk to Microsoft folks, although that was a nice fringe benefit.  I wanted to spend more time with people who had the same job, challenges, and duties that I had.  I wanted to get tips on how to do my job better, and that’s something I didn’t get from Microsoft.

But back to the press release, which notes:

“Based on Microsoft’s release cycle history (major release cycles run approximately every 3 years, with minor ones often coming in between), 2011 or 2012 will likely be a launch year. It would be disappointing for the community to lose out on the advantages of being in Seattle during a potential release year.”

Uh, hold on a second here.  If major release cycles run every 3 years, and minor ones come in between, then the odds of having a release on any given year is 2/3!  Every year is a potential release year.  As far as I’m concerned, if Microsoft wants to promote a release, they can come to the community.  This is exactly their strategy in Europe – they’re coordinating with community leaders to host regional events all over Europe for the launch of SQL Server 2008 R2.  Why does the community have to come begging to Microsoft’s doorstep?  They’re the ones making money off us.

The final word is that we’ll be in Seattle for the foreseeable future:

“The Board has decided to hold PASS Summit 2011 and PASS Summit 2012 in Seattle for the reasons listed above.”

Well, the good news is that they laid out their rationale in public for all to see.  That’s awesome, and I cheer their movement toward transparency.  I don’t agree with their decision, but I’m glad they didn’t try to hide the poll results from us.  The users clearly don’t agree with the leadership, but we elected ‘em.  I’m also excited to see that this little ruckus has caused at least one person, Jack Corbett, to consider running for election this year.  If you don’t like the way things go, this might be a great year to run.  Having 81% of the members on your side on a given issue makes for a great campaign platform, that’s for sure.  “My name is ___, and I’m committed to bringing the PASS Summit to the East Coast, unlike the current Board who wants to keep it in Seattle through 2012.”  Presto, you’re in.

The PASS Summit Location Survey Results is a post from: Brent Ozar - Too Much Information.
If you'd like to quote this content on other sites, please read how to use my blog's content.

Fusion-IO ioDrive

How fast are SSDs?  So fast that good ones overwhelm the capacity of the connection between the server and the drive.  The SATA bus maxes out at around 300 MB/s, and a good SSD can saturate that connection.  In order to get your money’s worth out of an SSD, you have to connect it with something faster than SATA.  Value is especially important given the pricing of solid state drives – more on that in a minute.

Fusion-IO ioDrives get around this limitation because they’re not SATA drives; they plug directly into your server’s much faster PCI Express bus.  These cards can push several times more data per second than SATA drives can. Other vendors using this approach include OCZ Z-drives and RAMSAN.  Of course, this connection method only pays off when the drive uses top-notch memory chips, and after briefly testing some of Fusion-IO’s products in their lab, I can vouch that they’re using the good stuff.

How Fast Are Fusion-IO Drives?

As a former SAN administrator, I’m anal retentive about reliability and uptime.  I’ve heard FusionIO drives sold as a “SAN in your hand,” but with a single drive, there’s not enough failure protection for my personal tastes.  I wouldn’t run any storage device without redundancy, so I ran most of my tests in a RAID 1 configuration – a mirrored pair of Fusion-IO SSDs.  Keep in mind that since these devices have their own built-in controllers, any RAID setup must be a software RAID setup managed by Windows.  Software RAID has a bad reputation, but it’s the only choice available when working with these drives.  I was initially worried about the performance impact of software RAID, but I didn’t have anything to worry about.

I tested several different ioDrive models using my SQLIO scripts as seen on SQLServerPedia and got blazing results.  Here’s a fairly typical set of results from a pass doing random reads:

C:\SQLIO>sqlio -kR -t2 -s120 -o8 -frandom -b64 -BH -LS P:\SQLIO\TestFile1.dat
sqlio v1.5.SG
using system counter for latency timings, 2929716 counts per second
2 threads reading for 120 secs from file P:\SQLIO\TestFile1.dat
using 64KB random IOs
enabling multiple I/Os per thread with 8 outstanding
buffering set to use hardware disk cache (but not file cache)
using current size: 10240 MB for file: P:\SQLIO\TestFile1.dat
initialization done
CUMULATIVE DATA:
throughput metrics:
IOs/sec: 22787.54
MBs/sec:  1424.22
latency metrics:
Min_Latency(ms): 0
Avg_Latency(ms): 0
Max_Latency(ms): 106
histogram:
ms: 0  1  2  3  4  5  6  7  8  9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24+
%: 100  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0

I know what you’re saying, though.  “Wow, Brent, that’s faster than any SAN I’ve ever seen before, but solid state drives are slower for writes, right?”  Yes, here’s a set of results for writes:

C:\SQLIO>sqlio -kW -t2 -s120 -o1 -frandom -b64 -BH -LS P:\SQLIO\TestFile1.dat
sqlio v1.5.SG
using system counter for latency timings, 2929716 counts per second
2 threads writing for 120 secs to file P:\SQLIO\TestFile1.dat
using 64KB random IOs
enabling multiple I/Os per thread with 1 outstanding
buffering set to use hardware disk cache (but not file cache)
using current size: 10240 MB for file: P:\SQLIO\TestFile1.dat
initialization done
CUMULATIVE DATA:
throughput metrics:
IOs/sec: 10114.24
MBs/sec:   632.14
latency metrics:
Min_Latency(ms): 0
Avg_Latency(ms): 0
Max_Latency(ms): 54
histogram:
ms: 0  1  2  3  4  5  6  7  8  9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24+
%: 100  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0  0

Write performance was about half as much as read performance, but that wasn’t because Fusion-IO drives were significantly slower in writing.  They managed the same 500-600 MBs/sec write performance when using one drive alone (rather than RAID).  When I put them in a RAID 0 (striped) configuration, I frequently got over 1,000 MBs/sec.  To put things in perspective, the fastest solid state drives in Anandtech’s recent showdown only achieved 337 MBs/sec in their absolute best case scenarios.

The only way to outperform a Fusion-IO drive is to invest six figures in a SAN and hire a really sharp SAN admin. These drives consistently outperformed every storage I’ve ever seen short of SANs wired up with true active-active multipathing software, because typical 4Gb fiber connections can’t sustain this kind of throughput.  Random access, sequential access, lots of threads, few threads, reads, writes, you name it – once we had them configured properly, I couldn’t get them to perform any slower than 550 MBs/sec, which is faster than you can drive 4Gb fiber.  Connecting directly to the PCI Express bus really pays off here, and makes everything simple.

Configuring SQL Server on a SAN is hard.  To really wring the best performance out of it, you have to design the pathing, the queue lengths on the host bus adapters, the RAID arrays, the SAN caching, the database filegroups and files, and sometimes SQL Server table partitioning in order to get everything to work in concert.  Fusion-IO drives return SQL Server configuration to its easiest: plug it in, put data and logs on the same array, and just create databases with the GUI, using one data file and one log file.  As long as your hardware environment is solid, it’s pretty darned hard to screw up a database configuration on Fusion-IO drives.

The Drawbacks of PCI Express Drives

Connecting data storage devices as PCI Express cards isn’t all unicorns and rainbows. Since they’re PCI Express cards, they can’t be used in blades unless you’ve got special PCI Express expansion blades.

These cards are connected directly to one server, which means they can’t be used in clustering environments.  If your data is mission-critical, you’re probably using a cluster to protect yourself from server failures.  Protecting from failures is even more important when you’ve got a single point of failure – like, well, a single PCI Express card with all your data.  If something should happen to go wrong with one of these devices, you can’t simply fail over to another server unless you’re doing synchronous database mirroring, something I rarely see in production.

You can get some form of protection by using software RAID 1: building a simple mirror with Windows between two Fusion-IO drives in the same server.  Whenever data is written to the volume, Windows will automatically write it to both Fusion-IO drives.  Software RAID gets a bad rap, but in my brief testing, I saw no performance penalty when using this configuration.

However, when a drive fails, you probably won’t be hot-swapping these while the server is in operation.  With server-class RAID arrays, you can pull a failed hard drive out of a server and replace it on the fly.  The RAID controller will rebuild the array while the server is still online.  Data access speeds will be slower while the array is rebuilt, but at least the server can stay up the entire time without an outage.  Not so with PCI Express cards: the server will have to be pulled out of the rack and opened up in order to access the drives.  This requires careful cabling – something I don’t see often in datacenters.

And Yes, You Need to Be Paranoid

During my testing, before Fusion-IO ironed out all of the configuration issues, I kept having drives drop offline.  Normally I’d blame my own stupidity, but my tests were run in FusionIO’s datacenter, on their servers, configured by their staff.  I connected via remote desktop, set up SQLIO per my SQLIO tutorial at SQLServerPedia, and ran the tests.  During my tests, ioDrives appeared to have failed and FusionIO staff had to replace them.  It took several weeks for us to narrow down several unfortunate problems.

If you truly try to push your IO subsystems to the limit, a Fusion-IO subsystem will expose more weaknesses than other storage subsystems because it has so much more throughput.  Some of the problems included motherboard issues, driver problems, OS configuration errors, and even insufficient power supplies that couldn’t handle the load of multiple drives.

Buyers need to be aware that this is a version 1 product with version 1 best practices and documentation.  When you put something like this into your infrastructure, make sure you’re actually adding reliability.  In my post about adding reliability to your infrastructure, I pointed out:

“The only way a twin-engine plane is more reliable is if just one of the two engines is enough to power the airplane safely. If the airplane requires both engines in order to maneuver and land, then the second engine didn’t add reliability: it just added complexity, expense and maintenance woes.”

The ironic part about my FusionIO testing woes was that they only happened in RAID scenarios.  The drives were actually more reliable without RAID – when I added RAID, I could knock a drive offline in minutes.  The faster we wanted to go, the more careful the team had to be with other pieces of the infrastructure.

FusionIO drives solve a real problem, and they can deliver staggering performance, but just like any other new technology, you should test them thoroughly in your own environment before deploying them in production.  Make sure to test them in the exact configuration you plan to deploy – if you’re going to deploy them in a RAID configuration, test them that way, rather than testing individual drives and assuming they’ll hold up in RAID configs.  In the case of Fusion-IO drives, you should probably even test using similar power supplies to production in order to improve your odds.

Where I’d Use Fusion-IO Drives in Database Servers

If you’re experiencing heavy load problems in TempDB, and if you’re not using a cluster, a Fusion-IO drive can probably solve the problem with less engineering effort than any other solution.  Simply shut down the server, drop in an ioDrive, change SQL Server’s TempDB location to point to the ioDrive, and start SQL Server up again.  Your TempDB IO won’t travel over the same storage paths that your database needs, which frees up more bandwidth for your data and log traffic.  TempDB requests stay on the PCI Express bus and don’t hit your SAN.

If you’ve got power, cooling, and space constraints in your datacenter, but you need to add more storage performance (but not necessarily capacity), a Fusion-IO drive makes sense.  You’ll gain more storage throughput this way than by adding several shelves of bulky, hot hard drives that require lots of power.  On a cost-per-gigabyte basis, this won’t make economic sense, but if you’re buying storage for performance needs, the cost equation is different.

If you need to scale an OLTP database that doesn’t require high availability, you might consider skipping clustering and trying database mirroring instead.  Use two SQL Servers in the same datacenter, both equipped with Fusion-IO drives, and make sure they’ve got plenty of bandwidth between them to keep up with synchronous (or asynchronous) mirroring.  You could argue that this provides a higher availability than clustering, since it uses two different storage back ends.  I’ve had SANs fail, and I can see how this might be attractive in some environments.  StackOverflow strikes me as an excellent candidate – the databases are small enough to fit on Fusion-IO’s drives, and the servers are rack-mounted as opposed to blades.

I’m also intrigued at the ioDrive’s potential to offload transaction log load.  I can envision a scenario where databases are in full recovery mode, but the load of logging and transaction log backups is starting to put a strain on the server’s IO.  Moving the transaction logs onto the Fusion-IO drive eases the load on the SAN (not just the hard drives, but the cabling between the servers and the SAN controller).

I wish I had SQL Server benchmarks for these scenarios to share with you, but the testing process ended up taking several weeks, and I ran out of time.  Thankfully, Paul Randal is blogging about his experiences with Fusion-IO drives.

My verdict: these drives can solve some tough challenges. I’m not saying that because I’m being paid to, either; Fusion-IO was gracious enough to give me access to their labs for my testing, but I didn’t get compensated.  Quite the opposite – I sank a lot of time into this project.  Folks who follow me on Twitter may remember that I struggled with Fusion-IO during the initial stages of the testing as we went through one hardware failure after another.  After the problems we encountered and the weeks of investigation, I’m glad to finally be able to say without hesitation that you should check out FusionIO’s products.  Their throughput may push the rest of your infrastructure to its limits, but hey – that can be a good problem to have!

This post is a part of T-SQL Tuesday, which is focusing on storage/IO this week.  If you liked this post, head over there to see what else is happening in the blogosphere this week!

Fusion-IO ioDrive Review is a post from: Brent Ozar - Too Much Information.
If you'd like to quote this content on other sites, please read how to use my blog's content.

You can access the videos, slide decks, links, scripts, and more for free at QuestKB.com/Live.

#SQLServerPediaTraining Archive Now Online is a post from: Brent Ozar - Too Much Information.
If you'd like to quote this content on other sites, please read how to use my blog's content.

“If it looks, smells and tastes like bacon, then it has to be #sqlserverpediatraining. Best training there is.“

I agree!  Here were some of the other quotes:

“The best tool to detect performance issues is to use a DMV (delicious meat vitamin) BACON. #sqlserverpediatraining” – @CameronMergel

“Quest has saved my bacon more than once. #sqlserverpediatraining” – @CameronMergel

“select * from #SQLServerPediaTraining” – @charlezhunt

Dr. Ozar MDF, Database Shrink

“you know it’s serious when Dr’s gloves come off. #sqlserverpediatraining #questqa http://tweetphoto.com/13147759” – @greeleygeek

“#sqlserverpediatraining Excellent training session, so dedicated they even drop test their schwag before sending it out” – @Tarwn

“I loved learning that @BrentO is @kekline ’s private dancer, dancing for cache #SQLServerpediatraining #questQA” – @SQLChicken

“when @kekline is in the crib, @BrentO drops a Kindle like it’s hot.. #SQLServerPediaTraining” – @SQLJackal

I want to thank everybody who made that training day so much fun.  Kevin, Ari, and I had a blast, and we’re already looking forward to the next one.

And the #SQLServerPediaTraining Netbook Winner Is…@JBrown414! is a post from: Brent Ozar - Too Much Information.
If you'd like to quote this content on other sites, please read how to use my blog's content.

My Lenovo, the Swedish Chef

Just tweet what you thought of our free SQL Server training event this week and include #SQLServerPediaTraining in your tweet.  We’ll pick up your tweets over at the SQLServerPedia stream, a new feature of SQLServerPedia that we’re testing out, and you can vote on what people are saying by clicking them up/down.

Saturday morning at 9am Eastern time, I’ll pick the highest-rated tweet by user votes and send them my Lenovo S10-2 that I picked up in Sweden during my epic European trip of doom.  It does have a US power supply (now), the extended 6-cell battery, and Windows 7 (not that starter version either), but it has a Swedish keyboard.  If you like, I’ll throw in a few stickers too!

The contest is open to US residents only.  (Sorry, but I’ve had a world of hurt every time I’ve tried to ship something to another country.)  Quest employees are not eligible.  I reserve the right to delete tweets that are inflammatory, spammatory, or anything else that doesn’t give me warm and fuzzy feelings.  I run contests with an iron hand and a Jello brain.

So go tweet with the #SQLServerPediaTraining hash tag, and then vote on your favorite tweets.  Good luck!

Wanna Win a Netbook? is a post from: Brent Ozar - Too Much Information.
If you'd like to quote this content on other sites, please read how to use my blog's content.

It’s true.  Wikipedia says so.  Sure, the side effects of methamphetamine are more than just anorexia and euphoria, but if you’re looking to be an upbeat, skinny person, meth gets the job done.  SQL Server has a couple of similar features that might make your queries faster today – but they’ll make you sing the blues tomorrow.

Forcing Index Use with Index Hints

If you’re not happy about the indexes SQL Server chooses to use when executing your query, you can show it who’s boss by forcing it to use specific indexes like this:

SELECT FirstName, LastName
FROM dbo.Employees WITH (INDEX (IX_Lname_Fname))
WHERE LastName IN ('Fritchey', 'Shaw')

The kids with Erector Sets were all jealous of Timmy.

This forces the SQL Server engine to use the index IX_Lname_Fname as part of the execution plan rather than looking at all of the available indexes and picking what it thinks is the best choice.  You’re telling SQL Server that you know more about the schema and the data than it does.

Whenever you hard-code an index into a query, you open up a can of performance tuning worms:

• If you add a better index, the engine won’t use it – you’ll be tied down to that one index until you touch every hard-coded query
• If that index gets dropped, your query will be in a world of hurt – you can’t consolidate indexes when you run into performance problems
• DBAs can’t do performance tuning without the developer recompiling the app
• When SQL Server is upgraded, it may be able to build a better query plan – but not if you forced a specific index

You might get a faster query today – but you can’t react as quickly to the problems of tomorrow.  The engine has another query plan trick, too.

Forcing Execution Plan Guides

Plan guides don’t just force SQL Server to use a specific index; the ambitious DBA can tweak all kinds of things about how the query is executed.  You can force things like table scans.

The nice thing about plan guides is that you can change them without recompiling your code and changing the queries.  The DBA can do performance tuning without getting developers involved every time.

Problem is, when you’re troubleshooting a query that isn’t behaving the way you’d expect, you can see index hints just by looking at the query.  Plan guides, on the other hand, are transparent – they’re happening inside the engine, behind the scenes, and you have to know to look for them.

But wait – it gets worse.  Plan guides exist at the server level.  When you’re troubleshooting the same query on development, QA, and production servers, and you’re getting wildly different results, do you stop to think if there’s a plan guide working behind the scenes?  Are your plan guides set the same across all three servers?  Are you sure?

SQL Server 2008 Query Performance Tuning Distilled

A Better Solution: Learn More About the Engine

Rather than telling SQL Server who’s boss, consider picking up Grant Fritchey’s excellent book, SQL Server Query Performance Tuning Distilled.  It explains:

• How statistics influence the accuracy of the execution plan
• How to build better, more selective indexes that the engine will be more likely to use
• How to understand if the execution plan is the problem, or just a symptom

Yes, we also cover this in our book SQL Server Internals and Troubleshooting, but if you’re the kind of person who’s considering index hints and plan guides in production, Grant’s book dives into more detail on these features.

2 Features of SQL Server You Should Avoid is a post from: Brent Ozar - Too Much Information.
If you'd like to quote this content on other sites, please read how to use my blog's content.

You may not know me, but I’m one of your readers.  You’ve become part of my routine.  I really like your stuff, but I’ve never really said anything until now.  I don’t want to leave a comment, because there’s not really a place in your blog to leave comments like this.  I know I could email you directly, but I’m too shy.

There’s a few things you could do that would really get me more involved in your blog.

Make it easy for me to leave a comment. Every now and then, I like (or dislike) one of your posts so much that I want to leave a comment. I want to just start typing right below the blog post.  If your blog says “anonymous comments are disabled” and directs me to sign up for an account to something, odds are I’m just going to close my browser and go on to the next blog.  You might have made the signup process really easy, but I’ve been through enough signups-from-hell that I don’t even want to bother anymore.

Respond to my comments. Most of my comments are questions to you, the author.  I’m hoping that you answer my question or respond to my opinion.  I’d be so excited if you took the time to answer back.  Even if I’m not leaving a question, just knowing that you acknowledged me gives me a little rush, because I’m new to this whole blogging thing, and bloggers seem like celebrities.

Email me when new comments are added. After I leave a comment, I don’t usually think to bookmark the page and come back to it later to see if anything’s changed.  If you’re using WordPress, all you have to do is install the plugin Subscribe to Comments, and I get a nice little checkbox when I leave a comment.  All new comments are emailed to me, and as soon as I see one, I’ll come back to your site again to keep the conversation going.  I’ll even start responding to other people’s comments, and next thing you know, your blog is a lively discussion.

(But not like that.)

If you write a series, make links between posts. If I happen to stumble across one of your posts in a series, I might want to read the other parts too.  In each post, include links to the other posts in the series.  The more posts I read, the more likely I am to leave a comment thanking you for your work.

Include the full post in your RSS feed. I know, you really want lots of hits on your blog.  You think that if you give me just a little taste in my RSS reader, I’m going to come to your blog to read the full article.  Sometimes I will – but sometimes I won’t.  That means I’m going to read less of your stuff.  If you insist on taking this approach, you had better put one hell of a lot of work into the first 2 sentences of each post, because that’s your only chance to capture my attention.

Before you give up on me, think differently. If you’re frustrated about why I’m not leaving comments, start approaching your blog differently.  Ask questions.  Take polls.  Say crazy things that challenge my complacency.  Beg me for help.  Start a meme.  Run a contest.  There’s a million things you can do to get me more involved, and for more inspiration, check out Problogger.

Believe me – I leave plenty of comments on other blogs.  The problem isn’t me – it’s you.

Dear Blog Author is a post from: Brent Ozar - Too Much Information.
If you'd like to quote this content on other sites, please read how to use my blog's content.

This weekend I splurged on the Headway WordPress theme, a premium theme that has a slick GUI editor.  I can make different pages look wildly different without too much effort.

Before

After

Here’s some of the changes so far:

Reorganized links across the top – I used to organize pages by popularity, but screw that.  I’m going for more logical organization now, so I’ve got links for About Me, Books, Consulting, SQL Server Articles, SQL Server Training Videos, Twitter, and Upcoming Events.  The Consulting and Videos sections are new, and I’ll be fleshing those out more over the coming weeks.  Yes, I still work for Quest full time, but I do consulting in my spare time.

Featured Video on the home page – I’m uploading more stuff onto YouTube now that they support high definition video.

Rotating “Featured Posts” and “Woohoo!” sections – as you watch the page, some parts of it rotate every few seconds.  I used to have logos for the Microsoft MVP program, Tom LaRock’s blog rankings, my book, etc, and now those all just rotate.  It’s slow enough that it’s not too obnoxious.

Magazine-style home page and SQL Articles page – I’ve got a lot of buried treasure in the site: articles people find via Google, but they don’t find when navigating through BrentOzar.com’s menus.  I’m starting to expose more of the popular articles through these browsable pages.  I’ll add thumbnail images for them, but probably not until the next version of Headway.  Headway 1.6 doesn’t pay attention to WordPress 2.9’s awesome new thumbnail functionality.

I’m still working on it, but I had to stop pouring effort into it and get back to my MCM studying.  The jury’s still out on Headway.  The visual designer loads and refreshes very slowly, it’s got a steep learning curve, and some things don’t seem to work the way I would expect.  It doesn’t play nice in any way, shape, or form with the DB Cache Reloaded plugin – Headway’s Visual Editor tool barfs all over itself when the cache is on.  I have to keep remembering to disable the cache when I’m editing the site, or else things keep exploding.

If you take the Headway plunge, back up your database first.  Headway says it uses the search engine meta tag information from the All-in-One SEO plugin, but that was not the case for me.  I appear to have lost all my SEO work on posts.  Not a big deal – I’ve got daily database backups – but I’m guessing not everyone is so fortunate.

Web Site Spring Cleaning is a post from: Brent Ozar - Too Much Information.
If you'd like to quote this content on other sites, please read how to use my blog's content.

Friday 2/26 – SQL Server Performance Tuning for Race Car Drivers
Nashville, TN and LiveMeeting

Times are tough even for the best drivers: Helio Castroneves is dancing for money and Danica Patrick is doing ads for what appears to be an adult services company.  Maybe it’s time to switch careers, and Brent has just the thing.  Use your hard-earned knowledge of high speeds, million-dollar hardware and surviving disastrous crashes to become a SQL Server performance tuner!

In this session, Brent will show you:

• Why Colin Chapman would check for indexes before adding new ones
• The importance of well-tested safety gear to performance tuning
• Why not monitoring your servers is like overdriving your headlights
• Just like races are lost in the pits, uptime records are lost during maintenance windows

Click here to launch the LiveMeeting at noon Central.  Audio will come through your computer speakers.

March 3rd: All-Day Free Virtual Event
Performance Tuning & Troubleshooting with DMVs

We’re doing an all-day live virtual conference on how to use dynamic management views (DMVs) to do SQL Server performance tuning and troubleshooting. The whole thing will be broadcast live in 720p, and we’ll be taking questions via chat and Twitter. We’re going to have sessions at the beginner, intermediate, and advanced levels taught by:

• Kevin Kline (Blog – Twitter) – Microsoft MVP since 2004, author of SQL in a Nutshell, founding board member of the Professional Association for SQL Server, and all around good guy.
• Ari Weil (Blog – Twitter) – Product Manager for Quest’s performance products, and knows way more about waits, performance tuning, and SQL Server architecture than anybody should.
• Brent Ozar (Blog – Twitter) – your humble author. Okay, well, author anyway.

Register now and just for registering, you’ll be entered into a contest for free goodies!

April 17 – SQLSaturday Chicago

SQLSaturday is a free day-long event with all kinds of community speakers. I’m excited to say I’ve applied to be one of the speakers! Kevin Kline and I will be doing the keynote, and I’m doing two other sessions:

BLITZ! 60 Minute Server Takeovers

You’re minding your own business in your corner office – well, no, you’re a DBA, so it’s just your cubicle – when somebody says, “Did you know about this SQL Server over here?” Suddenly, you have to find out what the server’s doing, how it was set up, and whether things are working correctly.

In this all-demo session, Brent will take over a SQL Server and show you:

• How to quickly assess a server’s health using a set of scripts
• How to catch dangerous development issues that will haunt you later
• How to quickly put some basic measures into place to minimize your risk

Perfmon and Profiler 101

These two tools are the key to successful performance tuning. I’ll show you how to get started with these tools, how to slice and dice the results, and even how to data mine the results to look for interesting trends. Whether you like to monitor performance with native tools or third party utilities, knowing how the native tools work will help you get better results out of your tools.

Attendees will learn:

• Why Perfmon is like a Nissan GT-R’s dashboard
• What Perfmon counters to measure, and what they mean
• Four common performance symptoms and how to cure the problems

Upcoming SQL Server Presentations is a post from: Brent Ozar - Too Much Information.
If you'd like to quote this content on other sites, please read how to use my blog's content.