DBA’s spend good amount of time monitoring and troubleshooting TempDB. This may include IO Bottleneck, Allocation Contention, DDL contention, Space issues, etc. IO bottleneck is a common issue. Workload demands increased over time and many DBA’s jump to conclusion that IO sub system needs to be upgraded. Yes, IO sub system upgrade may be a final solution but not the first option.
In my opinion, Memory pressure can be checked first. If you do not have enough physical memory on the box, extra IO can occur. You can check the following performance counters like Buffer Cache Hit Ratio, Page Life Expectancy, Chekpoint pages/sec, etc to see if there is memory pressure.
Second, you need to know which queries cause maximum IO. There may be throwaway queries causing IO contention or even well written queries but not utilizing the right indexes, etc. Following code snippet gives you top 10 queries that are generating maximum IO across all executions. If you find a culprit after investigating the execution plan, you can probably re-write the queries.
SELECT TOP 10
statement_start_offset as stmt_start_offset,
(SELECT SUBSTRING(text, statement_start_offset/2 + 1,
(CASE WHEN statement_end_offset = -1
THEN LEN(CONVERT(nvarchar(MAX),text)) * 2
END - statement_start_offset)/2)
FROM sys.dm_exec_sql_text(sql_handle)) AS query_text,
(total_logical_reads + total_logical_writes) DESC
Hope this helps