The place will be loaded with SQL Server all-stars, oh and I’ll also be there. Some will be Microsoft employees and others will be SQL Server MVP’s. Some will be local to the Redmond area while others are flying in from across the country.

It is a huge honor to be invited to speak on my favorite feature (AlwaysOn) of my favorite piece of software (SQL Server) at Microsoft’s world headquarters. I just arrived in Redmond and looking forward to a private speakers dinner tonight, as well as the public after party tomorrow night. Networking is one of my favorite activities since I enjoy meeting new geeks and talking about all sorts of dorky topics.

So if you are going to this event, please come up and introduce yourself. I’d love to get to know you. Also, Red Gate may be giving me a license key or two to give away, so ask me about it.

Microsoft SQL Server has had Mirroring starting with version 2005, and now that SQL Server 2012 is released, there is an additional feature set called AlwaysOn (marketing info | technical info). AlwaysOn in currently an Enterprise only feature, so Standard Edition users will still need to use Mirroring. I would describe AlwaysOn this way: it is a beautiful blend of Clustering and Mirroring smeared with Special Sauce. That may not be the best description, but being a fast food junkie, it really hits home with me. Don’t be afraid of the word Clustering though. Shared Storage (e.g. A SAN) is NOT required. AlwaysOn leverages some of the Clustering features to manage Quorum and the actual failover process.

When a failover event occurs, the failover process is usually quite snappy. Somewhere between 15 seconds and a minute usually. That said, there are a few reasons why failover may be on the slower side:

1. Virtual Log File Count (VLFs)
2. Cloud Disk Latency

1. High VLF Count Slows Failover Time

In simple terms, a Transaction Log file (.LDF) is made up of 3 or more segments called Virtual Log Files. Each segment can either be Active or not. The sizes are variable based on a formula related to the size of each growth. If SQL Server is set to autogrow and the tlog grows in small 1MB or 10% increments, there could be thousands of VLFs.

When SQL Server performs a failover, lots of transaction log entries need to be reconciled (read and write activities) in a process called Recovery (or Crash Recovery) so SQL Server can bring the database online in a healthy state (see ACID properties of a transactional database). Each VLF requires a little bit of overhead to process, so if there are thousands of VLFs, the process can be noticeably slower.

The solution is to properly shrink (which will deallocate the VLFs), and properly grow (which creates more VLF ‘segments’) the transaction log. The key term is properly. If done correctly, an 8GB transaction log file (.LDF) should contain between 19 and 35 VLFs. See this blog post from Kimberly Trip (b|t) for more details.

And you can check to see how many VLFs your database has by running this undocumented SQL Statement and counting the number of rows returned:

DBCC LogInfo

2. Cloud Disk Latency Slows Failover

Another reason that the failover can be slow, is if the IO Subsystem is pokey. This can be the case with a Direct Attach Storage (DAS) RAID 5 configuration which also uses slow drives. Or a local SAN using an iSCSI interface (large TCP/IP Overhead) combined with a budget 1Gbps pipe rather than 10Gbps. Of course, this is all relative to your servers IO demand (which can be greatly lightened without hardware upgrades by query/index tuning, etc.)

But for today, we are going to focus specifically on cloud storage. Amazon AWS EC2 (which we have a few clients using) is a prime example. True, Amazon has awesome data centers, and lots of computing power. However they also host Netflix and a bunch of other IO intensive companies. If the zillionth person decides to watch a new episode of Arrested Development (did you know it’s coming back?) and the data happens to be sitting on the same disk spindles as your database, you could have some latency issues. So if you are using EC2 or similar to house your replica, the failover will trigger a crash recovery on the EC2 replica to being it online. If this is timed while a lot of other IO unrelated to your account, then you may have a delay getting your server back up and operating.

This is known as the noisy-neighbor problem. And this can happen with any shared hosting, and even on your own corporate SAN (and even on local DAS drives housing multiple databases or other files). You don’t think your SAN has only your database on the underlying physical disks do you? Sure, your SAN Admin told you that this LUN is just for you, but they lie with the truth (and many times they just plain lie) by carving out your LUN from the same disk pool as a million other LUNs and [mis]leading you into believing that what you see on F: is the only data on those disks.

So if you are having slower than expected failovers, you can test the IO latency with a tool like IOMeter or similar.

The goal was to prepare some general technical people who don’t know a thing about databases/SQL Server and get them to the point where they can restore a database to a Test server, and write some queries so they can write some reports. So we entitled this class SQL101. Since the main goal was to have them effective at a basic level in just 4 hours, I organized the training in the following way:

• Break it into 50 minute chunks to allow frequent breaks and side-bar Q and A so everyone can keep up. No SQL-Child Left Behind!
• A few slides in PowerPoint to introduce a new topic and provide a little theory.
• A demo where they watch me perform the task we just discussed.
• A lab where they all do the same (with a twist) and I walk the class to provide assistance.
• Rinse and Repeat as necessary for each topic within each of the 4 parts.

I felt that this format would engage different parts of the brain in cycles, cover the theory needed to understand what they need to do (but in spoon fed sizes), and have them walk out of the class having performed each piece. I thought that the end result will be that people will exit the class and feel, not only that they CAN do it, but they already DID do it. I wanted them to rush back to their desk aching to crank out some joins using their new found skills. Therefore, all examples, demos, and lab exercises used the two databases that they will be making reports against so they can become accustomed to how they are normalized and where the fields of interest are.

The class was very much engaged and learned very well. Of course, some had a little experience walking into the class, and they were always asking about the content that happens to be on the next slide. I did provide some extra credit exercises on some of my labs since I figured this might happen (there’s always one).

They reported absolutely loving the class. One week later, my contact at EOG told me what great feedback he received from the class and that he personally said it was the best technical training he has ever attended. I bet he’s never attended any other class and has nothing to compare it too. Nonetheless, he asked me to begin working up a SQL 201 class to pick up where we left off. I don’t think they’ll be ready for Windowing functions in TSQL, but there is a lot more we can cover. Maybe I’ll give away Itzik’s book to prep them for the SQL 301 class they’ll likely ask for.

Most of the content for the SQL101 class is too basic for a SQL Server Users Group presentation. However, I may break it into 4 separate hour long presentations and provide it to a .NET users Group or two. I’m not sure if and how this format will work though, since it is about 35% lab, requiring everyone to have the same example databases on their own laptop. Maybe I should get some SQLWatchmen branded USB thumb drives with the presentation slides, demo scripts and sample databases to hand out.

Usually, I present this session in person to a live audience, which makes it very easy for me to have good audience contact and interact (bidirectionally) with the audience. I typically ask how many folks have used Mirroring in production and how many have used Clustering. The normal show of hands allows me to adapt my illustrations and comparisons, technical depth, and pace according to what I perceive the majority of my audience knows.

In this case, since it was a web presentation with muted attendees, I was a bit blind. This was not a problem for me though, since I typically present 3h webinars to audiences ranging from 30 to 400 attendees and I do this about 4 or 5 times a year as an Iron Speed MVP. Same drill here: PowerPoint, tons of demos, gotta have a bunch of corny jokes to keep folks awake, and everyone is muted. Meanwhile, I pretend they are all laughing [at me] while really they are sighing and probably checking their email. I use the chat feature to collect feedback and I use the Question and Answer feature to screen questions. It does make it more difficult to handle questions as compared to an in person presentation, but it is not a big deal.

My presentation to S3OLV was very well received. Charlie Jones, who runs S3OLV, said it was among the best presentations he’s seen and if I ever need a speaking reference, have them contact him. Wow, what a nice compliment! I think he must have gotten me confused with one of my favorite technical presenters Grant Fritchey (b|t). And no, I’m not kissing butt. I really enjoy his high energy active and sharp presenting skills.

Nathan Wimberley winner of Employee of the Quarter

SQLWatchmen company lunch was at GattiTown this quarter. As soon as we got in the door it was a mad dash to the pizza buffet. Everyone piled their plates high with pizza and we made our way to the cartoon room. After we finished eating and sipping soda an announcement was made that the company was going to begin implementing Employee of the Quarter. I also explained to the team that the winner of Employee of the Quarter would be someone that performed above what was expected of them on projects. Which in this office the team makes it very difficult to choose just one winner.

Congratulations Nathan Wimberley! For Q4 in 2012 Nathan Wimberley was announced as our Employee of the Quarter for his extraordinary execution on a very time sensitive project. Nathan was presented with a signed certificate, and a gift card. I also took this opportunity to commend the entire team for all their hard work in the year 2012. The team seemed very eager to see who will win Employee of the Quarter for Q1 in 2013!

Nathan dancing on Dance Revolution

Once we were filled with caffeine and pizza it was off to the game room. The company bought everyone token cards to play a few games. The team really enjoyed playing good old games such as, Skee Ball, basketball, and air hockey. Some of the many highlights of the day were the bumper cars, watching Nathan dance on the Dance Revolution game, and winning 400 tickets on a Deal or No Deal game of chance!! In the end we combined all our ticket earnings that totaled 1,750 tickets! It was difficult for the group to choose from all the prizes. We finally agreed upon a board game for a future company lunch, splat balls, mustaches, small trinkets, and candy. All in all the team had a blast!

It was refreshing to get out of the office for a few hours and just have fun with fellow team mates. These company lunches have proved to be a great way to show the office that they are appreciated, and helps us all form a closer bond so we are more effective when working together.

Make sure you check back to see where our next company lunch will be held, and who will win Employee of the Quarter next! Do you have company lunches too? Tell us about them!

Part of the team getting ready to burn rubber and smash into one another

Cotton Candy!

Skee Ball

Being a FoRG is open primarily to Microsoft SQL Server MVP’s and offers a lot of opportunity to help the community at large as well as Red Gate. I might participate in authoring or reviewing a book (Red Gate is a publisher of FREE ebooks as well as outstanding printed books), providing feedback on beta versions of their software and providing free license keys as give aways on their behalf as I travel the country and speak at SQL events. Oh, and did I mention that I get free licenses for all of their tools? I do need to know their tools if I’m going to review them, right? Of course, I already own many of their tools since they are so helpful.

I’m really looking forward to the next year of helping the SQL Server community in whatever capacity I am able.

Thanks Red Gate!

Red Gate SQL In The City (SITC) – Seattle

On November 5, 2012, I attended the Red Gate SQL In The City Seattle. It was fantastic free SQL training. I especially enjoyed the sessions from Adam Machanic (b|t), Brent Ozar (b|t) and Steve Jones (b|t). It was nice catching up with my many friends as well.

Kalen Delaney (b|t)

PASS Summit 2012 – Seattle

For the rest of the week, I was at the PASS Summit. Every year, this is the main event that I most look forward to. Thousands of SQL geeks fly in from around the world to attend, and the speakers are the best of the best.

I attended Bob Ward’s precon. Mind blower. If you are far past the basics of SQL Server, then this is an awesome session! Very technical. It was a 500 level precon so it was not for those with weak hearts or other medical conditions. And of course, he pulled out the debugger. We expected nothing less!

While I was there, I was sure to attend the annual User Group Chapter Leaders meeting. As a chapter leader running the ###CACTUSS Central SQL Server Users Group in Austin, TX, I find this a great help to learn how other cities and countries and overcoming the challenges inherent to running a User Group. We are also kept up to date on the latest methods of ordering books, emailing our registration list, finding and organizing speakers and so on.

If you have never attended the PASS Summit, you will want to ensure that you attend each year. It is the best training on the planet, and you can purchase a DVD of the sessions to continue your learning when back in the office. Honestly, I have many friends from around the world who pay international flights on their own dime if their company won’t pay. Yes, that is how valuable this event is. You don’t want to miss it.

So if you are in the U.S., the decision to attend this event is a no brainer. See you next year in North Carolina!

AlwaysOn is a kind of ‘super-mirroring’ for the Enterprise edition of SQL Server 2012. Mirroring is still supported in SQL Server 2012 and for the next couple of versions for those of you who have the Standard Edition of SQL Server 2012, but it was officially announced as deprecated so it will not be included in future versions. AlwaysOn is one of the most anticipated features and one which is causing a lot of buzz. I have implemented it for two of my customers since it was released a few months ago, and it is a lot of fun to work with.

Basically, it is like a cross between mirroring and clustering. But don’t be scared by the word ‘clustering’. AlwaysOn uses Clustering technologies for the Quorum voting and the failover logic, and DOES NOT require you to use shared storage (SAN). I repeat: you do not need a SAN to use AlwaysOn. Additionally, you can now have multiple ‘replicas’ (a fancy new word for primary or secondary), and you can configure your secondary replicas to be read only so you can offload your reporting/backup workload! Uh, you will need to buy more licenses for a secondary IF it will be used to perform ‘work’, like servicing reports or used as a backup source, but it will NOT NEED additional licenses if it is ONLY used as a passive node.

There are tons of other great features that we will break down and discuss like what an Availability Group is and details on how it works, what a VNN (Virtual Network Name) or Availability Group Listener is, Synchronous vs. Asynchronous, the flexible failover policy, and on and on. Nearly half of my presentation is a demo where we set it all up together and watch it fail from a real web application so you can see how it will work from your users perspective.

So if you are going to be in the Austin area on October 1st 2012, register now for this free training event, and I’ll see you there!

SQLSaturday 107 – Great Success

I was excited to go to SQLSaturday 107 for three main reasons. For one, my SQL Server 2012 AlwaysOn session was picked and I love giving that one. Secondly, Kalen Delaney (blog| @sqlqueen) was giving an all-day precon on SQL Server Execution Plans: Tuning And Management and I just had to see that! Lastly, I wanted to reunite with my #sqlfamily once again. I brought one of my employees Tim Smith (blog | @echoboombomb).  This was his first SQLSaturday.

I was also very pleased with how the event went, so thanks to the organizers, (Nancy Hidy-Wilson blog | @nancyhidywilson, Will Howard blog | @willsqlonline, Jonathan Gardner (blog | @jgardner04 and others) for their hard work! It was truly a delightful event.

SQLWatchmen was also a sponsor to provide a little financial help so the event would go well.

Arrival In Houston Thursday Evening

Upon arrival, Tim Smith and I had a pleasant dinner with Kendal Van Dyke (blog | @SQLDBA. That was arranged via Twitter, so if you are new to these events, get a twitter account and tune in, you just may end up having dinner with a PASS board member. So never underestimate the power of networking.

Precon w/ Kalen Delaney

Kalen Delaney was in Houston for the week training the ACE’s from Idera Software. While she was there, she extended her trip a little and taught an awesome precon titled SQL Server Execution Plans: Tuning And Management to provide folks with low cost, all day training going deep into her subject.

Someone Misstated Kalen and Twitter Went Crazy

She projected her voice well, so even the folks at the back of the room could hear her easily. Course, I was sitting at the table she was presenting from, but some ear plugs fixed that right up! Her manner and style was very pleasant, but most importantly, her content was highly technical and explained very methodically and clear. I loved it.

As the precon was about to start, we found that the projector VGA cable was not long enough to reach Kalen’s demo location. No problem, we quickly solved this by getting rid of 1 table, which seats two, and moved Kalen closer to the projector. The show went on, but with two fewer people: Tim and me. I dashed to a closed Radio Shack, pleaded with the manager an hour before they were officially open and he rung me up a VGA extension cable early! Great guy! I dashed back to the precon, with my ‘ticket’ to see the rest of the program in hand, and waited in the corridor for Kalen to take a break so we could rewire the projector, add in the missing table and provide us a place to sit. So SQLWatchmen became the proud sponsor of a VGA Cable.

While in the hallway outside the class, Nancy and I had a great discussion about, VGA cables, User Group stuff, event planning and financials, and life. It was very nice conversation.  Before long, it was break time and we rewired the projector and Tim and I got to see the rest of the presentation.

So there I was, sitting at the teachers desk feeling like I did in school when the teachers didn’t want me out of their sight. She said that I looked a little warm, but someone on twitter blew that way out of proportion. You can see a clip of that to the right.

A Message For All SQLSaturday Organizers

Here is one observation: most new people that I spoke with at the precon, had no idea what a SQLSaturday was and had no idea there was one the next day. Apparently, they’re bosses registered them for the precon (and apparently paid for them), forwarded the email and said: “Here, go to this training”. Three new people that I asked if they were going tomorrow, all said the same thing. So I have a call for SQLSaturday organizers: Promote the SQLSaturday several times AT THE PRECON. Your promotional emailings may go to the boss who registered, not the attendee. Maybe prepare a print-out about the following days’ SQLSaturday (hours, sessions, map, and how to register), and hand those out at the precon!  The SQLSaturday 107 team did make an announcement.

SQLSaturday 107 Sessions

The morning started with my session at 8:30 am: AlwaysOn – Finally A Usable ‘Mirror’ . The projector kept going into sleep mode while I was in the middle of using it. After some failed pre-session diagnostics, I dashed to my car in the parking lot to grab the projector that I needed to fail over to. I’m glad I brought it, but my students were even more happy.  Yes, I had a Disaster Recovery plan for the projector and was able to keep my presentation AlwaysOn (AlwaysOn is a trademark of Microsoft Corp, all rights reserved and all the rest of that stuff). After the initial projector hiccups, all went well. About 80% of the class were Clustering professionals, so I dove into some 300 material rather than just sticking to the basics.

William Assaf’s (blog | @william_a_dba) session SQL Admin Best Practices with DMV’s.
I really enjoy William’s style, what a great teacher. In this class, I learned that there is a DMV for collecting disk details starting in SQL Server 2008 R2. I did not know this. Awesome job, great mix of humor without going overboard, and a well polished presentation.

Christina Leo’s (blog | @christinaleo) session SQL Server Internals: It’s What’s for Dinner!
This was nice to see Christina’s presentation finally. We have presented at several of the same SQL Saturdays lately, but I have not been in her class as of yet. Boy, what excitement! She is so excited about what she teaches that her excitement bubbled over into the rest of the room. This fast-talking girl knows what she’s talking about and you’d better drink a Coke or have your coffee before entering this very technical and fast paced class. She covered SQL Server Internals like a pro. Beautiful slides with many animations as well. Very easy to follow.

Lunch

Lunch was full of fun. You might think that a simple meal would be uneventful, but around SQL geeks, you never know what to expect. In this case, a bunch of us were at a table, including Wes Brown (blog | @sqlserverio), Benjamin Nevarez (blog | @BenjaminNevarez), Argenis Fernandez (blog | @DBArgenis), Mike Fal (blog | @Mike_Fal) and others, when Jack Corbett (blog | @unclebiguns) walks up. The short jokes begin to fly, then the tall jokes (Jason Horner blog | @jasonhorner), then the skinny jokes, then the fat jokes. It got pretty brutal but we all laughed a ton. No hard feelings of course. And in the end, we came up with a new invention for Jack so people can find him when he is in a crowd: A belt with an orange bicycle flag attached to the back so the flag sticks way up in the air.

Robert Davis’s (blog | @SQLSoldier) session Replication Magic: Initializing From Backups.
I volunteered to be a room monitor for this session to help ensure things go smooth for Robert. This was an awesome presentation where he showed us how to start Replication from a backup. Nice slides, great demos and expert handling of the questions.

Kalen Delaney Taught An Awesome Precon

Vicky Harp’s (blog | @vickyharp) session Edge Case Testing for the Database Professional.
This was a session that I wanted to see for a while. In fact, Vicky and I have spoken at the same SQL Saturdays IN THE SAME TIME SLOT at every SQLSaturday that we both spoke at.  Austin #97, same time slot. CO Springs #104, same. This was a treat and I really enjoyed the material and the discussion. Very interactive session.

I wanted to see Kalen Delaney’s session Costs of Concurrency, but I figured it was packed (which I heard later that it was). So I took the opportunity to speak with the various sponsors about their products, community, thanking them for sponsoring this event. I also got to talking with Mark Rosenberg (blog | @m_rosenberg) from New Horizons in Austin and found that he is very excited about what we are doing in Austin and is looking forward to sponsoring our own SQLSaturday in September.

Conclusion

I had a fantastic time at this SQLSaturday, caught up with a lot of friends and learned a great deal as well. If you have never been to a SQLSaturday event, I encourage you to find one within a few hours driving distance and GO! You won’t regret it. Experts, book authors, MVP’s and MCM’s speak at these and it is free to attend thanks to the sponsors! You just can’t loose.

P.S. – Why SQLWatchmen Sponsored

I was asked by several folks at the event where they should put their raffle ticket for SQLWatchmen since I did not set up a table/booth, and didn’t even set up a box to collect the tickets. The answer is simple: SQLWatchmen did not sponsor this event to grow the business, we sponsored to help the community and assist Nancy and company with pulling off a great event. SQLWatchmen focuses on helping small companies who do not need a full time DBA, but which run SQL Server for their critical systems, like Sharepoint, Dynamics, iMIS, etc. So nearly all attendees (DBA’s), along with their employers (who already have a DBA) are not likely to be in need of our help. So SQLWatchmen was not a sponsor in an effort to find new customers, but simply to help the community and to give a little something back to the community that I fell in love with some time ago.

Sounds strange, I know, but that is why there was no booth. This is the same reason I sponsored last years Austin SQLSaturday, and the same reason I’ll do it again this year.

The blog has been quiet over the Winter, so where have I been lately?

Simply put: On the road with the SQL Server community! I’ve been doing a lot in the community lately, as well as having our busiest Winter ever with SQLWatchmen.  So instead of making a bunch of blog posts about what I’ve been up to, I decided to just make one big itinerary.

SQLFamily at a Speakers Dinner

Over the past few months, I have:
1) Presented at the CACTUSS Central SQL Server Users Group in Austin on the topic of AlwaysOn. This was on September 20, 2011.
2) Co-Coordinated SQLSaturday 97 in Austin on throughout August and September.
3) Spoke at SQLSaturday 97 In Austin on AlwaysOn using CTP3 on October 1, 2011.
4) Attended the PASS Summit in Seattle for the first time on the week of October 10, 2011.
5) Attended the annual Iron Speed MVP Summit in San Jose and met with fellow MVPs from around the world. This was the week after the PASS Summit (week of October 17, 2011).  I actually flew straight from Seattle to San Jose and didn’t return back to the office for two weeks straight.  That was rough.
6) In November and December, I prepared for and presented a three hour webinar each month on C# development with SQL Server.
7) On January 7, 2012, I flew to Colorado Springs to give the AlwaysOn presentation there. They put me in the Auditorium and there was probably about 60 people in attendance there, but I didn’t count.
8) Went on a traveling road trip, speaking throughout the middle of the state of Texas presenting my AlwaysOn session. The first stop was at the SALSSA SQL Server Users Group in San Antonio on January 18, 2012.
9) Then the following day, drove up to Dallas and spoke at the NTSSUG SQL Server Users Group to a large and enthusiastic audience of 83 people! That is a big users group.  That was on January 19, 2012.
10) On March 10, 2012, I was in Tampa, FL at SQL Saturday 110 where I enjoyed presenting my AlwaysOn session.
11) On April 16, 2012, I presented my AlwaysOn session to the CACTUSS Central SQL Server Users Group.
12) And I just returned from SQL Saturday #107 in Houston where I presented my AlwaysOn presentation on April 21, 2012.

I think that is a lot of community activity and traveling, so my blogging suffered a bit lately. I know, not an excuse. I’ll get better at not being a perfectionist so I can ‘just blog’. I’ll also get better at not making excuses.

On top of the above ‘external’ activities, I have also been busy with ‘internal’, or less visible community activities:
13) Running a SQL Server Users Group (CACTUSS Central meetings). Well, this one is kinda visible.
14) Sitting on the board of POSSE.
15) Going through the election process with POSSE where I won a 2 year board of directors position and continue my role as the Treasurer.
16) Beginning the early planning stages for our 2012 Austin SQL Saturday (Finding venues, finding sponsors and deciding who should be on this years planning committee).

Well, no one has ever accused me to lacking passion for the community.

17) Oh ya. Almost forgot, I’ve been running a DBA consulting company, where we have been experiencing a lot of growth requiring that I hire two new staff members in Q1 2012. We also moved into a larger office location on February 24, 2012.

Is that why I haven’t been sleeping much lately?!?