NEWID vs NEWSEQUENTIALID decides whether new uniqueidentifier keys land at random in your clustered index or in order at the end. This post loads 20,000 rows with each default, measures pages, fullness and fragmentation, and shows the limits of NEWSEQUENTIALID. It also answers what happens when you switch the default on a table that already has rows.
Check Index Fragmentation With Row Count in SQL Server
To check index fragmentation with row count, read sys.dm_db_index_physical_stats in SAMPLED mode and keep the page count and row count next to each percentage. This post builds a demo with three tables, lists every index, filters out the tiny ones, loops over every online database, and rebuilds the one index that matters.
How to Know If Index is Forced on Query? – Interview Question of the Week #246
Question: How to Know If Index is Forced on Query?
Answer: Let me tell you a brief story so you understand the context of the question.
Columnstore Indexes Without Aggregation: Do They Still Help?
Columnstore indexes without aggregation still help when a query reads many rows but only a few columns. This post builds a rowstore table and a columnstore copy of one million rows and compares reads, size and execution mode. It also shows the case where a plain b-tree index wins.
DBCC DBREINDEX: Replace It With ALTER INDEX in SQL Server
DBCC DBREINDEX still runs on SQL Server 2025, but it is deprecated, and ALTER INDEX does everything it did and more. This post rebuilds a fragmented demo table both ways, shows the options only ALTER INDEX has, and finds old uses of DBCC DBREINDEX in your own modules, Agent jobs and the deprecation counter.
Unique Index Performance: What a UNIQUE Index Saves
Unique index performance gains come from what the optimizer learns, not from smaller pages. A tested demo compares pages, reads, plans and CPU for a plain index and a unique index. It also shows the error a duplicate causes.
Indexes and SELECT Queries: A Coffee Mug, Tested
Indexes and SELECT queries met on a coffee mug from my friend Harvey Powell. Its two sides say indexes slow down SELECT queries and to drop unused indexes. The tests here show when a forced index reads 46 times more pages, what the usage counters report for an unused index, and why zero reads did not mean zero purpose.