troubleshoot locking / blocking in sql server

exec dbo.sp_WhoIsActive

–Filters–Both inclusive and exclusive
–Set either filter to ” to disable
–Valid filter types are: session, program, database, login, and host
–Session is a session ID, and either 0 or ” can be used to indicate “all” sessions
–All other filter types support % or _ as wildcards
@filter = ”,
@filter_type = ‘session’,
@not_filter = ”,
@not_filter_type = ‘session’,

–Retrieve data about the calling session?
@show_own_spid = 0,

–Retrieve data about system sessions?
@show_system_spids = 0,

–Controls how sleeping SPIDs are handled, based on the idea of levels of interest
–0 does not pull any sleeping SPIDs
–1 pulls only those sleeping SPIDs that also have an open transaction
–2 pulls all sleeping SPIDs
@show_sleeping_spids = 0,

–If 1, gets the full stored procedure or running batch, when available
–If 0, gets only the actual statement that is currently running in the batch or procedure
@get_full_inner_text = 0,

–Get associated query plans for running tasks, if available
–If @get_plans = 1, gets the plan based on the request’s statement offset
–If @get_plans = 2, gets the entire plan based on the request’s plan_handle
@get_plans = 1,

–Get the associated outer ad hoc query or stored procedure call, if available
@get_outer_command = 1,

–Enables pulling transaction log write info and transaction duration
@get_transaction_info = 1,

–Get information on active tasks, based on three interest levels
–Level 0 does not pull any task-related information
–Level 1 is a lightweight mode that pulls the top non-CXPACKET wait, giving preference to blockers
–Level 2 pulls all available task-based metrics, including:
–number of active tasks, current wait stats, physical I/O, context switches, and blocker information
@get_task_info = 1,

–Gets associated locks for each request, aggregated in an XML format
@get_locks = 1,

–Get average time for past runs of an active query
–(based on the combination of plan handle, sql handle, and offset)
@get_avg_time = 0,

–Get additional non-performance-related information about the session or request
–text_size, language, date_format, date_first, quoted_identifier, arithabort, ansi_null_dflt_on,
–ansi_defaults, ansi_warnings, ansi_padding, ansi_nulls, concat_null_yields_null,
–transaction_isolation_level, lock_timeout, deadlock_priority, row_count, command_type

–If a SQL Agent job is running, an subnode called agent_info will be populated with some or all of
–the following: job_id, job_name, step_id, step_name, msdb_query_error (in the event of an error)

–If @get_task_info is set to 2 and a lock wait is detected, a subnode called block_info will be
–populated with some or all of the following: lock_type, database_name, object_id, file_id, hobt_id,
–applock_hash, metadata_resource, metadata_class_id, object_name, schema_name
@get_additional_info = 1,

–Walk the blocking chain and count the number of
–total SPIDs blocked all the way down by a given session
–Also enables task_info Level 1, if @get_task_info is set to 0
@find_block_leaders = 1,

–Pull deltas on various metrics
–Interval in seconds to wait before doing the second data pull
@delta_interval = 0,

–List of desired output columns, in desired order
–Note that the final output will be the intersection of all enabled features and all
–columns in the list. Therefore, only columns associated with enabled features will
–actually appear in the output. Likewise, removing columns from this list may effectively
–disable features, even if they are turned on

–Each element in this list must be one of the valid output column names. Names must be
–delimited by square brackets. White space, formatting, and additional characters are
–allowed, as long as the list contains exact matches of delimited valid column names.
@output_column_list = ‘[dd%][session_id][sql_text][sql_command][login_name][wait_info][tasks][tran_log%][cpu%][temp%][block%][reads%][writes%][context%][physical%][query_plan][locks][%]’,

–Column(s) by which to sort output, optionally with sort directions.
–Valid column choices:
–session_id, physical_io, reads, physical_reads, writes, tempdb_allocations,
–tempdb_current, CPU, context_switches, used_memory, physical_io_delta,
–reads_delta, physical_reads_delta, writes_delta, tempdb_allocations_delta,
–tempdb_current_delta, CPU_delta, context_switches_delta, used_memory_delta,
–tasks, tran_start_time, open_tran_count, blocking_session_id, blocked_session_count,
–percent_complete, host_name, login_name, database_name, start_time, login_time

–Note that column names in the list must be bracket-delimited. Commas and/or white
–space are not required.
@sort_order = ‘[start_time] ASC’,

–Formats some of the output columns in a more “human readable” form
–0 disables outfput format
–1 formats the output for variable-width fonts
–2 formats the output for fixed-width fonts
@format_output = 1,

–If set to a non-blank value, the script will attempt to insert into the specified
–destination table. Please note that the script will not verify that the table exists,
–or that it has the correct schema, before doing the insert.
–Table can be specified in one, two, or three-part format
@destination_table = ”,

–If set to 1, no data collection will happen and no result set will be returned; instead,
–a CREATE TABLE statement will be returned via the @schema parameter, which will match
–the schema of the result set that would be returned by using the same collection of the
–rest of the parameters. The CREATE TABLE statement will have a placeholder token of
–<table_name> in place of an actual table name.
@return_schema = 0

Datavault, Anchor modeling or Inmon 3NF

This article will give you the pros and cons of different datawarehouse architectures, so that you can choose which one is best for your customer. As a BI architect, my job is to be objective. Try not to defend the things you know, but try to keep an open mind.

We will limit our scope to 3 popular architectures: Datavault, Anchor modeling and Inmon 3NF.

Example domain: Nasa Facilities. This is just a simple listing of Nasa centers and facilities. Each Centers hosts one or more facilities.

Powered by Socrata

Modelled in Data Vault 2.0


MODELLED in Inmon style 3NF



  1. Comparisons between modelling techniques – Hybrid (including Data Vault) Roelant V.

Multiple schemas (MS)

Suppose we have 2 source systems A and B. Both systems contain information regarding employees. System A contains login_name and email address, System B contains department.

In our datawarehouse we like to merge this information. My preferred way of doing is is the following.

System A has a schema called A in My_DWH.

System B has a schema called B in My_DWH.

So in this example My_DWH contains the table A.user and B.user.
We can also create a table or view called dbo.user in My_DWH that contains a merged view of the two.

It’s important to have A.user and B.user for lineage. You need to be able to go back to the source of records in dbo.user.

* It’s better to use this pattern also when you only have System A in your data warehouse (i.e. there is only 1 user entity). Otherwise you will have a lot to do when System B is added at a later time.

When using this pattern in Staging, we avoid issues when different sources have the same table names.


When in the above example records from schema A and B are merged in My_DWH.dbo.user, we might get uniqueness issues of natural keys. To solve these issues we always add a column named src_schema_id to the natural key.

e.g. dbo.user
sur_key , src_schema_id          , natural_key ,   synonym_id 
1              , A                                   , ‘bvdberg’        , null
2              , B                                   , ‘bvdberg’        , 1

The combination of src_schema_id and natural_key is unique and we can use the [Syn] pattern to de-duplicate these records in DWH. Note that we can choose to only show sur_key 1 in the datamart. (surrogate foreign keys will not point to entities that have synonyms).


min and max function tsql

Author: Bas van den Berg
Create date: 2014-02-25
Description: returns the minimum of two numbers
select dbo.udf_max(1,2)
select dbo.udf_max(null,2)
select dbo.udf_max(2,null)
select dbo.udf_max(2,3)
ALTER FUNCTION [dbo].[udf_max]
@a sql_variant,
@b sql_variant

RETURNS sql_variant
if @a is null or @b > @a
return @b
if @b is null or @a > @b
return @a
return null

uthor: Bas van den Berg
Create date: 2014-02-25
Description: returns the minimum of two numbers
select dbo.min(1,2)
select dbo.min(null,2)
select dbo.min(2,null)
select dbo.min(2,3)
ALTER FUNCTION [dbo].[udf_min]
@a sql_variant,
@b sql_variant

RETURNS sql_variant
if @a is null or @b < @a
return @b
if @b is null or @a < @b
return @a
return null