Dig Deep!
?
Dig Deep
Root Cause That Bad Boy!
I first used that term back in 2006 or 2007. I was working for a Managed Hosting company. We hosted over 700 SQL Servers for different clients. Some of them paid for us to maintain, do backups, etc. Some did not.? We didn’t do any performance tuning…strictly installs, patching and break-fix.
But not just break-fix and forget, because that’s rude.
Break-fix and root cause, because fixing the same problem over and over is annoying to me, the client and everyone tracking tickets.? Its great for KPIs I guess, if you are graded on tickets closed, first touch resolution, etc. But that is level 1 support. We were level 2. I upgraded us to level 2.5 before I knew there was such a thing as ITIL and defined layers of support.
Quite simply, my team took the approach of “Hey look…a problem! Lets fix it and root cause that bad boy so it never happens again!”
It still amazes me how many problems we found that we could resolve and then sprinkle that fix over hundreds of other servers proactively. We cut the ticket count by 70-80 percent, and as a DBA or two left my team we saved their salary (one went to Microsoft, the other went to the Oracle team).
How does this apply today?
Exactly the same. We root cause everything we see here.
I’ve been trying to prove out a replication/application problem for quite some time and finally got it this morning!
My client has a 3rd party application that uses ad hoc and prepared statements…not stored procs. It’s a black box to me and will not be changed.
They use Transactional Replication to send everything they can to the Reporting server. Huge reporting/Analytics team,
At least once a week the Replication gets so far behind, the Analysts can’t work.
Check out this one table:
Do you see it?
Classic data amplification. Could be triggers, cascades or any number of other things…all inside that black box.
Since there are not stored procs, Replication is doing these writes one row at a time.
How would you solve this issue, without being able to touch the application or its indexes??
领英推荐
I have already proposed my solution.?
My recent LinkedIn post(s):
I’m doing my first online event! (Register here…it’s a Zoom meeting)
Interesting stuff I read this week:
Dude Perfect raises more than $100 million from Highmount Capital to hire CEO, expand business (axios.com)
Not much else interested me this week ???
Coming soon…an offer for new clients of Dallas DBAs!
SQL tidBITs:
The columns that make up your Clustered Index are automatically part of every non-clustered index. You don’t need to add them.
Sponsored by:
?Share with your friends…6,770 subscribers are ahead of them!
unconventional problem solver, hands-on expert (as required), transforming tech teams, companies, systems and processes.
11 个月Its amazing how many companies don't require an RCA for outages, interruptions or incidents. I trained my teams to write them with two distinct audiences -> internal section and external section. And external was the only information, verbatim, that could leave the company. Internal had a more complete analysis, warts and all, timelines and what we have done to prevent reoccurrence. The sooner we got those out the sooner everyone got the correct information and calmed down. Goes a long way towards building and maintaining trust.
SQL DBA | Azure SQL | Service now | AWS | Ex-TCS | Ex-Amazonian | Ex-Wiproite |
11 个月I'm curious