About a year ago, (almost to the date) my wife and I decided that we would knock a hole in the awl that separates our living room from the stairs that go up to the second floor of our house.  When we bought the house, it was originally a duplex that we had thought about renting half out.  Those ideas had changed when we decided that we would try to start a family.  By the way, our due date is tomorrow.

When we cut into the wall, we soon discovered something that most older house probably have; a nest of wires with no explanation as to what was what.  In addition to this, I also needed to re-route everything into a new workbox and consolidate everything that was duplicated between the two floors such as the shared porch light.  While I was at it, I thought I’d also hook up a set of lights four our walkway and backyard and also an outdoor receptacle.

So, as mentioned before, my wife and I are expecting tomorrow and with the added work of getting prepared for that, it took a little longer than I expected to complete.  Anyway, I got everything done except re-wiring the 3-way switch for the stairs leading to the second floor.  The porch, the receptacle, the lights outside all were completed with relatively no problems compared to this last switch.

I started working on this switch last week and couldn’t figure out how to get the wiring on this to work as the previous owners decided that they would not only not hook up the switch as intended but also decided to splice it into another wire on the other end.  I tried every conceivable combination of connecting these wires while popping the breaker a few times, or so I thought.  Since I has disconnected these wires over a year ago, I couldn’t just remember how these were connected originally.  However, when I first got into the wall, I remembered that the wiring looked so messy, I had to take a picture.  Actually I took 5 pictures (2 close-ups and 3 wide angle) which ended up saving me from a last minutes melt down.

So, from this story, the lesson to be learned is when you’re working on something, whether it be automotive, house or any other type of repairs, TAKE PICTURES!!!  They may end up saving you, I know it saved me.

When I started working in the computer industry, our company had only 6 servers realistically.  Sure, there were a few other servers that were always up and running but weren’t really ever used for day to day operations.  Also, there was a payroll machine in the back that ran some piece of Payroll software as was connected to a few other computers on their own token ring network.  But for the company to truly operate, there were only 6 servers.

In addition to the minuscule numbers of servers, there were a number of tasks that we (the computer department and then called CSS) were responsible to perform daily.  Since our company operated mostly overnight, our business day actually started at 12:00PM (Noon).  This meant that every day (M-F), someone needed to run a process to reset the system for the new days work.  Also, there were a number of reports that needed run daily,weekly, monthly and since we didn’t have e-mail, we were also responsible for printing and stuffing these reports into envelopes that would get delivered to various people though out the company either in our corporate office or at remote locations.  And if someone called off or had a vacation day that we weren’t aware of, there was sure to be someone complaining about not getting their report on time or at all.

In addition to the process to reset the systems and the daily reports, there were also numerous other jobs that needed run.  Luckily, I only had to work the late shift (that was responsible for 6 pages of tasks) a few times.

Now I look around today to see how things have changed.  We now have more than 100 servers, a few SANS, and an actual Data Center around which our still fairly new Corporate Building is centered.  The days of having to perform tasks manually at certain times is gone.  With new technology, we are able to schedule jobs and put in additional measures to alert for failure in the event that occurs.  I look at my DB Servers and see the 100′s of jobs out there that rarely fail and if they do, they know to wait a few minutes and try again.  I also look at our reporting environment where we now have the ability to create a report one day, schedule it the next day and never really have to worry about it again.  The users get their reports virtually at the same time everyday.  I don’t get called into peoples’ offices to discuss why a report went out late or why someone didn’t get a report at all and I don’t have to listen to a Texas Instruments 8920 Dot Matrix printer printing for 8 hours straight.  Oh how far we’ve come in a relatively short amount of time.

As miraculous it is how far we’ve come in my history at this company,  I also remember the work that happened in between the 2 times listed above as well as the hours/days/weeks/months/years of training that went into making us smart to be able to perform the work that was needed.  It is amazing and overwhelming to quantify the amount of effort that has gone into making our computer enterprise what it is today.

Lastly, I thought of the people who are no longer here.  My friends who have come and gone through the years; Those that I’ve successfully kept in touch with and those with whom I’ve lost touch.  It’s sad that I worked with these people for years and now they’re nowhere to be seen day to day.  However, in light of this, if everything needed to happen as it did to get to this point, I’d have to say it was worth it.

On a personal note, my wife and I are expecting our son in about 11 weeks.  So anytime not spent working has been used to focus on getting ready for our son to come home.  To date, we’ve gotten most of the nursery furniture together and painted the nursery.  My mother-in-law and I worked together to surprise my wife with a surprise shower in FL.  In a few weeks we’ll be having another one a little closer to home.

Let me step back a bit.  At work, I use my Dell Laptop running Windows 7 and various other OS’s via VMWare.  Most of these being Windows Server of one sort or another.  At home, my primary computer is my MacBook.  I find that using a Mac at home is any easy way to separate work computer usage from personal computer usage.

With that being said, let us return to speaking about Linux.  In the same way that PC vs. Mac separates work from home, Linus adds another separation for learning to use MySQL.  While I could install MySQL on any of these platforms, I find that when I have a dedicated VM with Linux installed along with MySQL, I don’t get distracted as much and can easily focus and learn this new DB platform.

I’ll have more to post on MySQL later.

Session #1 – Performance Tuning Made Easy – Thomas Larock

In this session, Larock discussed the methods that he uses when tuning a SQL Server for performance.  Using his “12 Step” method of trouble shooting and the 6 Sigma approach he showed how one could use the same exact method to resolve issues time and time again.  After discussing this for a while, he showed an example where he presented a case where a user would hand over a SQL script they’ve created that is performing poorly and expect you, the DBA, to fix it.  Up until now, I’ve had my own methods which seemed to match up with the majority of the session’s attendees, but after seeing his methods in action, I may be a changed DBA.  This was one session for whichI will definitely download the slide deck.

Session #2 – Raise the Bar by Writing Professional Database Code - Aaron Lowe

In this session, Aaron started the discussion about writing professional code and some of the practices he exercises when writing code.  He also explained that in addition to it just being a good practice, it can also be used as a method of future proofing one’s code.  Writing Professional Code is something I always try to do, so much so that even when writing adhoc queries to look into an issue or give someone a quick answer, I automatically use indentation with out even thinking.  Also, I’ve seen the downfalls, first hand, when some doesn’t use or believe in using strong data types in their code.  With all of this being said, Aaron showed some other practices that I don’t currently include in my coding.  Some of these included ending statements with a semi-colon.  While this is not necessary, future versions of T-SQL may require it now, by including it now when creating code, you ensure that you won’t need to re-visit your code if/when this requirement comes to fruition.  Another method was to unset anything you set such as nocount statements.  As is my own practice, I include nocount statements in my own code when I don’t need to count the number of rows affected.  However, I never set nocount back on when I’m done.  This topic started a heated conversation between one of the attendees and the speaker as to whether this was truly reverting your setting back to their original state or just setting them to the opposite state when completing your code.  The discussion was centered around calling one procedure from another and whether setting the nocount off would trickle up to the outside calling procedure.  This is one that I among other attendees will test come Monday morning.

I started writing this post a few weeks back right after attending SQL Saturday #67.  Since then I’ve been busy and hadn’t had chance to complete this.  I was about to delete this thinking it was too long after the event to finish this but then thought there’s no harm in completing it and I’d let the readers decide whether or not they wanted to read.  Anyway here is the continuation…

Session #3 – Baselines First, Troubleshooting Second – Erin Stellato

In this session, Erin (an avid panelist for Women In Technology) talked about the importance of baselines so that when disaster strikes you can have a clear comparison to show whats not performing as well as normal.  Also, the benefit of these sessions is that we are freely able to talk the truth about users and their perceptions.  Keep in mind this isn’t a slam session on users but she spoke of something that, being a DBA for 14yrs, I’ve come across more than once.  The scenario is that a user calls and says that the database and/or system is slow and then when you ask what is slow, they respond by saying everything or if you ask how much slower the system is, they respond with some number times slower.  The other problem, other than the user not being able to answer your questions is if they do answer them.  At this point, you have no way of validating their response, however, if one has baselines in place, then you would have a reliable way to compare the current system performance to the baseline.  Click on the link above to see the agenda and download the slide deck.

Session #4 - SQL Server Storage: Filegroups, Files, Partitions – Brent Ozar

This was probably the session I wanted to attend most.  I had seen Brent online a few times and listened to some of the pod casts from SQLServerpedia.com and found that in addition to being very knowledgeable DBA, he is also very entertaining.  This session didn’t disappoint.  In this session, Brent talked about getting the most performance out of your disks by using multiple Filegroups, File and Partitions.  Also, he talked a good deal on how to get the most out of a SAN environments by utilizing multipathing to increase the IO capabilities and throughput to your SAN.

Session #5 – Who Needs DMVs? You Do! – Tim Ford

In this, the final session of the day, we discussed the importance of DMVs (Dynamic Management Views) and DMOs (Dynamic Management Objects).  A re-occurring theme to these sessions is that one can use the tools already included with SQL to manage, troubleshoot and diagnose issues with SQL Server.  Among these tools are DMVs and DMOs.  Using these, you can very detailed information about the SQL Server installation as well as the server its installed on.  Another topic that keeps popping up is the use of wait stats.  Using the DMVs, you can very easily analyze these wait stats and determine what exactly is causing your pains.  In previous versions of SQL, some of this information was available thru extended stored procedures but the amount of detail that is now available is unrival by any previous version of SQL Server.

Conclusion

All in all, this was another very successful SQL Saturday from the perspective of an attendee.  While I was there, I was able to network with a few fellow DBAs and able to get a lot of information regarding SQL.  The more of these events I attend, I find myself willing to start conversations with people I don’t know.  Its comforting to know that most if not every person there is a fellow DBA who is a lot like me and enjoys working with and having discussions about SQL Server.

Regardless, I’ll be able to catch them when they’re posted for viewing in a few weeks.  If you have a chance to catch any of these online, do yourself a favor and watch them.  Also check out some of the older sessions SQLPass hosts.

When I was in High School, there 2 classes I enjoyed more than any other.  First, Music/Band/Choir was my overall favorite so much se that I actually started college as a music major with a focus on teaching.  Second, Math was another favorite because of the practice aspect of it.  More precise, I enjoyed Algebra 2/Trig.  With most of my classes, there was just a ton of reading with no application.  However, with music and math, application was the home work.

With that being said, a few weeks ago I faced an issue where I would need to convert a ton of information from a base 36 value to decimal and then back to base 36.  If you don’t know what base 36 is, it is the numbering where each column can contain 36 values (0-9,a-z) and each column to the left is a set of thirty-six.  Think of this as hexadecimal but instead of 16 values there are 36.

The reason for this request was that the transportation company I work for works with a partner to ship into Canada and their Postal Codes are alpha numeric and contain values 0-9,a-z.  Furthermore their postal codes are based on street and blocks as opposed to the US where ours are determined by zones.  From this Canada has exponentially more Postal Codes than the US so the only way we store these codes is in ranges.  Well, someone requested that we split up these ranges into individual Postal Codes.

In order to do this, I would first need to take the low and high values of the range and convert them from base 36 to decimal and then take every value in between and convert those from decimal to base 36.

As I began to write this, I thought “How can I make this more versatile?”  I decided that instead of writing a set of functions to convert a value from decimal to base 36 and back, I would instead write a set of functions that would allow a user to pick their base and supply the value to be converted and that’s what I did.  Running the script below creates a table and 2 functions.  The table is used to map decimal values to base 36 values.  The functions allow you to specify the base and value to convert.

I’ve gone through and documented the script and functions to hopefully answer any questions one may have.  However, if you do have a question or idea for refinement, post a comment and let me know.

Conversion

A couple of night ago, I got a call from work stating that a core system was running slow.  Multiple apps were being affected and no major changes had happened so this was quite inexplicable.

Normally I would look at the results from an sp_who2 to see if anything was locking up the DB.  Also, I would look at the server hosting the DB to see what the resources looked like.  The normal suspects included (CPU, Memory, Disk IO)

All of this was done and unfortunately, nothing showed up as be blatantly offensive.  A few months back I found a simple script that would provide some alternative and more detailed results to the normal sp_who and sp_lock commands.  Since then, I’ve modified the script time and time again until I had 3 different scripts.  One to return top active CPU, Memory, and IO sessions.  Most recently I’ve consolidate these 3 scripts down to 1.

With out further ado, here it is…

s0_edba_Now

Save this file and change to extension on .txt or .sql  If you execute this proc with no params you’ll get a usage message.  Between the use of the 2 parameters, you can select to sort the results by CPU/Memory/IO and also choose to see all active processes or just the top 10.

By using this proc, I was able to determine that there was a specific user that was always hovering at the top of the IO list and furthermore, I was able to see what that SPID was executing.

I found that the procedure that was being executed was trying to perform a delete of data older that 14 days old every time an insert was being done on a table.  Furthermore, the delete was filtering on a field with no indexes.

There have been numerous times when I see a DB getting very large and I need to create a purge or archive plan to keep the size of the DB under control.  With this script, I get a quick look at what tables are the largest and will give me the biggest bang for the buck.

Also the “Unused” information is a quick way to see where fragmenting may be occurring.

s0_edba_get_spaceused