February 12, 2010 at 4:24 pm
Brings up an interesting point. I sometimes quote large texts from others to respond to and put a long dashed line beneath.
I should have done a better job of pointing out my comments....!!
the "
" IFCode shortcut on the left hand side of the "post reply" text editor works wonders for clarifying that you're quoting a previous post.
have a good weekend
Craig Outcalt
February 18, 2010 at 9:42 am
This is a very good article indeed. Something every new DBA or would-be DBA should read and take to heart.
The only thing that doesn't come across clearly is just how difficult it can sometimes be to follow those tips.
It really can be very difficult to do even one of deliver CIA, avoid being a hero, and set definable expectations (let alone do all three). This is especially true if being DBA is something you do in your (non-existent) spare time from your "real" job, more so if the head of development doesn't see why developers should pay any attention to a DBA, more so again if the CEO firmly believes that there's no real need for a DBA (and anyway even if he wanted a DBA he couldn't hire one because the company is a start-up and the burn rate is already far too high), and yet more so still if the CEO believes that he (despite never having used a database or written a line of code) is better at estimating the time and effort needed to do something technical than anyone with any technical knowledge could possibly be -all situations that many in this forum will have come across (although they'd have to be really unlucky to come across all those problems at the same time).
(edit to sort out some typos)
Tom
February 18, 2010 at 10:58 am
Tom.Thomson (2/18/2010)
more so if the head of development doesn't see why developers should pay any attention to a DBA.(edit to sort out some typos)
This is where the DBA can live or die by email. Keep upper management (The Dept head's boss) informed by using blind copy (bcc) if very important stuff like this is going on, so when disaster does eventually strike you can fully document it. Be discreet and choosy about what you decide to email to upper management, yet still getting the crucial points across. I had this happen as well, and I fully documented it, and the department head ended up losing his job, instead of me. 🙂 E-Mail is a DBA's best friend, and I see too many that don't use it to their benefit by keeping people informed of what is going on with the company's multi-million dollar databases. Use it, it might save your job one day. Don't let the CTO tell you one day "If only you would have told me about this before, I could have done something about it." Let them fall on the sword, make sure it is not you. Email can really help a lot with this, but remember, to be discreet. HTH 🙂
"Technology is a weird thing. It brings you great gifts with one hand, and it stabs you in the back with the other. ...:-D"
February 21, 2010 at 11:16 am
Excellent article!
The only thing I would add is "getting used to being blamed". 😛 Network sucks, it’s DBA’s fault. SAN collapse, it’s DBA’s fault. Programmer updated data without a WHERE clause, it’s DBA’s fault….blah..blah..blah
Something else I liked to share: I was a programmer before becoming a DBA. I thought I knew how to program a database-driven application…..I was wrong. I didn’t know anything about indexes or how to write good queries. I wrote whatever that would gave me results then moved on. Yeah I was stupid and ignorance, but just learned lately that I’m not alone. There are still new programmers out there making the same mistakes I made, and I had to support their databases. Yikes! Talk about Karma, eh?
One question as a newbie: Do you need to know about database design and/or application in order to do performance tuning? I know about Profiler and PerfMon, but I wonder if these tools are enough to say these are good/bad queries without any knowledge of applications?
Regards,
P.S forgive my poor English.
February 24, 2010 at 6:09 am
I would say that you don't need in-depth application knowledge to tune SQL performance at all.
You do need to figure out why the query is written the way it is, and that to some degree involves a basic amount of business knowledge. IF you can read T-SQL and understand database design, you can figure out the business as you go by asking questions and not making assumptions.
You definitely need to know how to read T-SQL, run profiler (if you don't know what the problem is ) and know how to read a query plan and use IO statistics to see where the hard-hitters are.
The first thing I usually do is re-write the T-SQL trying to minimize problem areas in the plan.
Basically I try to reduce the number of rows searched and looped through... maybe take a sort out of the plan or an aggregation.
If all that fails, I'll add an index or add TO an index, or maybe even use an indexed view in a pinch.
Thanks for reading!
Craig Outcalt
February 24, 2010 at 6:31 am
Tom.Thomson (2/18/2010)
This is a very good article indeed. Something every new DBA or would-be DBA should read and take to heart.The only thing that doesn't come across clearly is just how difficult it can sometimes be to follow those tips.
It really can be very difficult to do even one of deliver CIA, avoid being a hero, and set definable expectations (let alone do all three). This is especially true if being DBA is something you do in your (non-existent) spare time from your "real" job, more so if the head of development doesn't see why developers should pay any attention to a DBA, more so again if the CEO firmly believes that there's no real need for a DBA (and anyway even if he wanted a DBA he couldn't hire one because the company is a start-up and the burn rate is already far too high), and yet more so still if the CEO believes that he (despite never having used a database or written a line of code) is better at estimating the time and effort needed to do something technical than anyone with any technical knowledge could possibly be -all situations that many in this forum will have come across (although they'd have to be really unlucky to come across all those problems at the same time).
(edit to sort out some typos)
All great points, Tom.
I think it fits nicely into the "we've all done things wer're not proud of" category where the fight for survival trumps the higher level functions.
I often think of of Abraham Maslow's Hierarchy of Needs
http://en.wikipedia.org/wiki/Maslow's_hierarchy_of_needs
and the capability maturity model in these situations.
Levels of the Capability Maturity Model
1. Initial (chaotic, ad hoc, individual heroics) - the starting point for use of a new process.
2. Managed - the process is managed according to the metrics described in the Defined stage.
3. Defined - the process is defined/confirmed as a standard business process, and decomposed to levels 0, 1 and 2 (the latter being Work Instructions).
4. Quantitatively managed
5. Optimized - process management includes deliberate process optimization/improvemen
Really, organizational maturity is not a tough sell to anyone who doesn't require a rectal craniotomy.
But we've all worked for that guy and had to do things that we really weren't proud of...
Here again, good communication and unwaivering belief that you're doing the right things can maybe set things moving in the right direction. One thing is for certain, organizations CAN and DO remain chaotic and ad-hoc under bad management.
Craig Outcalt
Viewing 6 posts - 61 through 65 (of 65 total)
You must be logged in to reply to this topic. Login to reply