Advanced SQL Server Transaction Log Analysis (Udemy.com)

Straight to the point, hand to hand learning and very practical, this is what an SQL course should look like

Great course to the point sweet and short. I loved it. Needed to quickly learn how to navigate the transaction log and this course taught me that.

A very light and superficial course. Sometimes it is even wrong: truncate is not heavily logs, in your case, it creates 12 lines but it is a fixed amount of lines so if you delete 1bn rows it is very minimal. No explanation for the .bak result despite the fact that there is one...

A very succinct introduction to the SQL Server transaction log. A must for anyone wishing to understand the function of the transaction log and learn how to explore its contents within the space of an hour or two.

This is a decent introduction to poking around the transaction log, and is worth the small time investment. There are a few inaccuracies, though. First, in lecture 9, the explanation provided for the absence of LOP_BEGIN_XACT records after a checkpoint is that they were "hardened" or "written to the log", which is incorrect. fn_dblog does show hardened log records. The trick is that by default the log reader only shows the active portion of the log (this can be changed with a trace flag as indicated by Paul Randal at https://www.sqlskills.com/blogs/paul/using-fn_dblog-fn_dump_dblog-and-restoring-with-stopbeforemark-to-an-lsn/). In quiz 2, a question asks how many LOP_BEGIN_XACT records will be produced by inserting 5 rows. This question does not fully determine an answer. If the 5 rows are inserted one by one as autocommitted transactions (as in the video), then the desired answer of 5 is correct. However, if the 5 rows are inserted as part of a single transaction (either because all 5 rows are inserted by a single statement or because multiple insert statements occur in a single transaction) then there will be only 1. The 5 rows could really be straightforwardly inserted as any number of transactions between 1 and 5. Finally, the description of intent locks is misleading at best. Intent locks are not some sort of pre-lock that suggests maybe we will take a lock later and maybe we won't. An intent locks is an actual lock that is requested and granted and can conflict with other requested and granted locks. The "intent" is not that maybe a real lock will be taken later, but that a more restrictive non-intent lock will be taken on an object lower in the lock hierarchy (e.g., when a row is deleted SQL Server might choose to take very real IX locks on the table and page and an X lock on the row). This allows SQL Server to take the more restrictive locks at the most granular level feasible, thus maintaining the desired level of isolation while allowing more concurrency than if non-intent locks had to be taken at every level in the lock hierarchy. Overall, this course does a fairly good job of introducing fn_dblog and fn_dump_dblog, but gets some details wrong along the way.

Useful information for beginners on Transaction Log Analysis. I would not say this is an Advanced presentation, since it merely introduces the topic.

The course material is directly applicable to the real world which I really liked. I really liked the section on 'whodunnit' as this is something I've never known how to do before.

I learned a great deal of information that I've never seen presented before. Very good presentation with real-world lab examples.

Now I know who dropped my table so I can beat them with a big stick! On a serious note whilst this course was short it was jam packed with useful examples that are applicable to real world scenarios, I learnt may nifty things that will definitely make life easier and I will be using them often. Thanks again Mike for another awesome course!

Mike has produced a great course that fully explains the SQL transaction process in a clear and concise method.