Sql server lesson12

System Monitor
 System Monitor, commonly referred to as PERFMON, is a
Microsoft Windows utility that allows to capture statistical
information about the hardware environment, operating system,
and any applications that expose properties and counters.
Perfmon.exe

System Monitor
 System Monitor is used to capture numeric statistics about
hardware and software components.
 It uses a polling architecture to capture and log numeric data
exposed by applications.
 An administrator chooses the counters to capture for analysis and
the interval to gather data.
 System Monitor then uses the definition supplied for the counters
and polling interval to gather only the counters desired on the
interval defined.
 Capturing one counter or 100 counters does not affect system
performance.

 System Monitor (continue…)
 Counters are organized into a three-level hierarchy: Counter
object, Counter, and Counter instance.
• Counter Object : is a component, application, or subsystem
within an application, must have at least one counter.
For example : Processor, Network Interface, and SQLServer :
Databases are all objects.
• Counter
o One or more counters are specified within an object,
and every object has to have at least one counter.
o You have counters for active transactions, data file size,
transactions/sec, and the percent of the transaction log
space currently in use.
o A counter can have zero or more instances.
• Counter Instance.

 The System object has a Processor Queue Length counter
that does not have any instances .
 The counter that captures the percentage of the log space
used within the SQLServer : Databases object has an
instance for each database as well as a cumulative instance
named _Total.
 When you define counters to capture, you can specify criteria
at any level within the hierarchy.
 You can capture an entire object, So System Monitor gathers
data for every counter within the object , and for each instance
available within the counter.
 Also You can capture data for a subset of counters within an
object for a subset of instances within a counter.
 System Monitor (continue…)

 Capturing Counter Logs
 To capture data for analysis, you need to configure a counter log,
which runs in the background when no one is logged on to the
machine.

 Capturing Counter Logs (continue…)
 Viewing counters in the graphical interface provided by System
Monitor is useful only for looking at the immediate state of a system.
 When setting up counter logs, it is recommended that you select
counter objects instead of individual counters to ensure you have
captured everything necessary for analysis.
 The sample interval determines how frequently Windows gathers
data for the counters specified in the log.
The default setting is every 15 seconds
To analyze a problem you want to decrease the polling interval.

 Capturing Counter Logs (continue…)
 You should always configure a counter log to run under a specific
account with sufficient permissions to access the applications for which
you are gathering counters.
 Most common causes of a counter log failing to start:
• Failure to define a specific account .
• Password expiration
• A locked out account
• A deactivated account

 Performance Counters
 There is a small set of common counters that can be used to
troubleshoot a variety of problems:
 System: Processor Queue Length
 Network Interface: Output Queue Length
 Physical Disk: Avg. Disk Queue Length
 An individual counter is generally used in conjunction with other
counters plus additional information about your environment to
diagnose a problem.
 Individual counters and groups of counters can direct you to an
area of the system that might need further investigation but
does not directly indicate a problem on the system.

 Performance Counters
 When the processor, network interface, or disk becomes
overwhelmed with activity, processes need to wait for resources
to be freed up.
 Each thread that has to wait for a resource increments the
corresponding queue length counter.
 For Example, a processor queue length of eight
indicates that there is insufficient processor
capacity on the machine and that eight requests
are waiting in a queue for a processor core to
become available
 The queue length can be greater than zero for very short
durations, having any queue length greater than zero on a routine
or extended basis means that you have a hardware BottleNeck
that affects application performance.

 Working With The SQL Server Profiler
 SQL Server is built upon an event subsystem, SQL Trace, which
is exposed for administrators to capture information associated to
over 200 events that can occur within an instance.
 SQL Server Profiler is only a GUI designed to work with another
feature of SQL Server called SQL Trace.
 SQL Server Profiler used to diagnose errors and improve the
performance and stability of your applications.
 You can use the data captured to monitor an instance for errors
or concurrency issues.
 You can also use Profiler to capture data that is used to optimize
the performance of queries being executed against the
environment.

 SQL Trace is a very simple tool. Its job is just to capture SQL Server-
related communication between a client and SQL Server.
 SQL Trace acts similarly to a specialized network sniffer that captures
traffic on the network related to SQL Server and allows you to see
exactly which events are being sent from the client to SQL Server.

 HOW CAN HELP DBAs
 Monitor the activity of the:
o SQL Server Database Engine
o Analysis Services
o Integration Services
• Monitor and identify performance-related problems with front-end application,
queries, T-SQL, transactions, and much more .
• Perform query analysis of execution plans
• Monitor and identify SQL Server errors and warnings
• Capture and replay traces of activity to replicate problems.
• Audit user activity.
• Analyze trace results by grouping or aggregating them .
• Create your own custom traces, and save them, for use whenever you wish
• Correlate Profiler trace events with System Monitor performance counters to
identify What event(s) are hogging server resources
• Capture data for analysis in the Database Engine Tuning Wizard
• Save trace results to a file, a database table, or to XML for later analysis
• Trace Profiler events programmatically, instead of using only the Profiler GU

 Defining a Trace
 You can define a trace using Transact-SQL (T-SQL), it is more common
to use SQL Server Profiler to define the trace.
 Also, you can start SQL Server Profiler from the SQL Server 2008
Performance Tools menu.
OR

 Defining a Trace
 SQL Server Profiler is an application that allows you to define
graphically settings that are translated into stored procedure calls to
create and manage traces.
 By default, when a trace is started using Profiler, all the events appear
in a grid within the interface, You can additionally save the trace data
to a table, a file, or both.
 Profiler can capture a very large number of events in a short amount
of time and when loaded into the grid within Profiler can require the
use of a large amount of memory.
 You can specify an upper size limit on the trace file to keep
the file from growing out of control
Enable File Rollover
After a trace file has reached the maximum size,
the file is closed, and a new file is opened.
Disable File Rollover
Profiler stops capturing events after the
maximum file size has been reached

Template Purpose
Blank An empty trace; allows you to create an entire trace from scratch.
SP_Counts Captures each stored procedure executed so that you can determine how many of each
procedure is being executed
Standard The most common template to start with; captures stored procedure and ad hoc SQL
being executed along with performance statistics for
each procedure and batch. Every login and logout is also captured
TSQL Captures a list of all the stored procedures and ad hoc SQL batches that are executed,
but does not include any performance statistics
TSQL_Duration Captures the duration of every stored procedure and ad hoc SQL batch
that is executed
TSQL_Grouped Captures every login and logout along with the stored procedures and
ad hoc SQL batches that are executed. Includes information to identify
the application and user executing the request, but does not include
any performance data
TSQL_Locks Captures blocking and deadlock information such as blocked processes,
deadlock chains, deadlock graphs, lock escalation, and lock timeouts.
This template also captures every stored procedure, each command
within a stored procedure, and every ad hoc SQL request.
TSQL_Replay Captures the stored procedures and ad hoc SQL batches executed
against the instance in a format that allows you to replay the trace
against a test system. This template is commonly used to perform load
and regression tests
TSQL_SPs Captures performance data for all ad hoc SQL batches, stored
procedures, and each statement inside a stored procedure. Every login
and logout is also captured.
Tuning Captures basic performance information for ad hoc SQL batches, stored
procedures, and each statement inside a stored procedure.
 Profiler ships with several templates that have events, data columns, and filters already
defined, so you can use them as a starting point.
 The following lists the general purpose of each template:

 Specifying Trace Events
 SQL Trace exposes more than 200 events that can be captured.
 Selecting the set of events is the most important action that you define
when configuring a trace .
 Events are classified into 21 event groups. You need to know which
events are used to solve various problems
 The most commonly used event groups are Locks, Performance,
Security Audit, Stored Procedures, and TSQL.
 The Stored Procedure and TSQL event groups are commonly captured
along with events from the Performance group to baseline and
troubleshoot query performance.
 The Security Audit event group is used to define auditing quickly across
a variety of security events, although the new audit specification feature
provides more secure and flexible auditing capabilities.
 Events from the Locks event group are commonly used to troubleshoot
concurrency issues.

 Specifying Trace Events
 Most events return a small amount of data, some can have a significant
payload on a very busy instance.
 The events to be very careful with are:
• Performance| Showplan *
• Stored Procedures| SP:StmtCompleted
• Stored Procedures | SP:StmtStarting
• TSQL| StmtCompleted
• TSQL| StmtStarting
 Before I ran my test code, I ran the DBCC FREEPROCCACHE command in
order to remove all currently existing query plans from SQL Server cache.
 You need to know which events are used to solve various problems.
For Example, resolving deadlocks, blocking, or stored
procedure performance.
 You can capture performance data on the entire stored procedure with
the RPC:Completed event and an entire batch with the
SQL:BatchCompleted event.

Event group PURPOSE
Broker 13 events for Service Broker messages, queues, and conversations
CLR 1 event for the loading of a Common Language Runtime (CLR) assembly
Cursors 7 events for the creation, access, and disposal of cursors
Database 6 events for data/log file grow/shrink as well as Database Mirroring state changes
Deprecation 2 events to notify when a deprecated feature is used within the instance
Errors and Warnings
16 events for errors, warnings, and information messages being logged. Events to detect
suspect pages, blocked processes, and missing column statistics.
Full Text 3 events to track the progress of a full text index crawl
Locks 9 events for lock acquisition, escalation, release, and deadlocks
OLEDB 5 events for distributed queries and remote stored procedure calls
Objects 3 events that track when an object is created, altered, or dropped
Performance
14 events that allow you to capture show plans, use of plan guides, and parallelism. This event
group also allows you to capture full text Queries
Progress Report 1 event for online index creation progress
Query Notifications 4 events to track the parameters, subscriptions, and templates for query notifications
Scans 2 events to track when a table or index is scanned
Security Audit
44 events to track the use of permissions, impersonation, changes to security objects,
management actions are taken on objects, start/stop of an instance, and backup/restore of a
database.
Server 3 events for mounting a tape, change to the server memory, and closing a trace file
Sessions
3 events for existing connections when the trace starts as well as tracking the execution of
logon triggers and resource governor classifier functions
Stored Procedures
12 events for the execution of a stored procedure, cache usage, recompilation, and statements
within a stored procedure
Transactions 13 events for the begin, save, commit, and rollback of transactions
TSQL
9 events for the execution of ad hoc T-SQL or XQuery calls. Events for an entire SQL batch as well
as each statement within a batch
User Configurable 10 events that you can configure with SQL Trace
 SQL Trace Event Groups

 After you have determined which events you want to capture, you then
need to determine which columns of data to return.
 Not all data columns are valid for all trace events.
 When you are trying to baseline performance, you can capture the following
events:
 RPC:Starting: fires when a stored procedure is first executed as a
remote procedure call.
 RPC:Completed: fires after a stored procedure is executed as a remote
procedure call.
 SQL:BatchStarting: fired whenever a new Transact-SQL batch begins.
This can include a batch inside or outside a stored procedure
 SQL:BatchCompleted: occurs when a Transact-SQL statement
completes, whether the Transact-SQL statement is inside or outside a
stored procedure.
 It is very unlikely that you would trace the SQL:BatchStarting or
RPC:Starting event because they cannot provide any performance
data necessary for a baseline or to troubleshoot performance issues.
 Selecting Data Columns

 The ApplicationName, NTUserName, LoginName, ClientProcessID, SPID,
HostName, LoginSID, NTDomainName, and SessionLoginName provide
context for who is executing a command and where the command is
originating from.
 Some useful column names:
 The SessionLoginName always displays the name of the login that was
used to connect to the SQL Server instance.
 The LoginName column accounts for any EXECUTE AS statements and
displays the current user context for the command.
 The ApplicationName is empty unless the application property of the
connection string has been set when a connection attempt is made to the
instance.
 The NTUserName and NTDomainName reflect the Windows account for
the connection, regardless of whether the connection used a Windows or
SQL Server login to connect.
 The HostName is particularly useful in finding rogue processes in an
environment because it lists the name of the machine that a command
originated from.
 Selecting Data Columns (continue…)

 The StartTime and EndTime columns record the time boundaries for an
event. The StartTime and EndTime columns are especially useful when
you need to correlate trace data with information from other systems.
 The ObjectName, ServerName, and DatabaseName columns are useful
for analysis operations.
 By including the ServerName and DatabaseName columns, you can easily
separate trace data between multiple instances/databases while still only
needing a single table for storage.
 A common practice is to have a single instance within your environment
where you import all your traces.
For Example: The ObjectName column for the RPC:Completed
event lists the name of the stored procedure executed so that
you can easily locate all calls to a specific stored procedure
that might be causing problems in your environment.
 Some useful column names:
 Selecting Data Columns (continue…)

 To target trace data, you can add one or more filters to a trace.
 A filter is essentially a WHERE clause that is applied to the event
data returned by the SQL Trace API.
 Filters are defined on data columns and allow you to specify multiple
criteria to be applied.
 Apply filter upon data Type of columns :
 Character data allow filters to be defined on a text string using
LIKE or NOT LIKE that can contain one or more wildcard
characters.
 Time-based data columns allow you to specify greater than or
less than.
 Numeric-based data columns allow you to specify equals, not
equal to, greater than or equal, and less than or equal.
 Binary data columns cannot be filtered.
 Applying Filters

 You can start, stop, and pause a trace.
 After a trace has been started the SQL Trace API return events
that match the trace definition while discarding any events that
do not match trace filter criteria.
 When a trace is stopped, all event collection terminates, and if
the trace is subsequently started again, all previous trace data is
cleared from the Profiler display.
 If you want to suspend data collection temporarily, you can
pause a trace.
 Upon resumption of a trace, subsequent events are appended to
the bottom of the Profiler display.
 Managing Traces
 Running Trace Automatically :
 By running traces via code, you eliminate all the overhead of a graphical
tool while also providing a means of unattended tracing through the use of
jobs within SQL Server Agent

 The trace modules that ship with SQL Server 2008 are:
 Managing Traces (continue…)
Name SP/Function Description
sp_trace_create This SP defines and create a new trace
sp_trace_generateevent This SP is used to create a user-defined event
sp_trace_setevent Used to add or remove events being traced
from existing trace and objects capture.
sp_trace_setfilter Includes and excludes data.
sp_trace_setstatus Starts, stops and closes a trace.
fn_trace_geteventinfo A function that returns the events and data
columns being captured by a trace.
fn_trace_getfilterinfo Return filter information associated with a
specific trace.
fn_trace_getinfo Returns information about a specific trace or
all traces.
fn_trace_gettable A function that reads one or more trace files
and returns the contents as a result set.

 Correlating Performance and Monitoring data :
 SQL Trace is used to capture information that occurs within a SQL Server
instance.
 System Monitor is used to capture performance counters that provide a
picture of the hardware resources and other components running on the
server.
 SQL Server cannot run without access to hardware resources, So The
state of hardware resources affects how well a server running SQL Server
functions.
 You might be able to diagnose an issue using only System Monitor or
SQL Trace.
 For Example, a query might be running slowly, but Profiler
tells you only how slowly a query is running.
However by adding in performance counters, you might find
that the reason queries are running slowly is because you
have insufficient processing resources.

 Correlating Performance and Monitoring data :
 Profiler allows you to view a trace file with a counter log.
 You can correlate a counter log with a trace file only if you have captured
the StartTime data column in the trace.

 Diagnosing Database Failures
 Error and informational messages related to your server running SQL
Server can be found in:
 Windows Event logs.
 SQL Server error logs.
 SQL Server Agent logs.
 Database mail logs.
 Windows Event logs are commonly viewed using the ((Windows
Event Viewer ))
Event Viewer
 The SQL Server error log is a text file on disk that can be opened and
viewed by any text editor such as Notepad.
 The current SQL Server error log is named errorlog (without any file
extension) and is located in the MSSQL10.<instance>MSSQLLOG
directory.
 You can also retrieve the contents of the current error log as a result set
by executing the system extended stored procedure
sys.xp_readerrorlog
 SQL SERVER LOGS

 Information that affects your SQL Server installation can be found in
three different event logs:
 System event log Contains system error and information messages
primarily related to hardware and operating system components.
 Application event logs The primary source of SQL Server
information that contains all the error and informational messages for
an instance, including service start/stop/pause messages.
 Security event log If you have enabled auditing of login/logout
events, each successful connect and disconnect from an instance is
logged in this event log
 SQL SERVER LOGS (continue…)

 The main purpose of the Error Log is to log error messages such as
database corruption, stack dumps, insufficient resources, and hardware
failures.
 The SQL Server Error log contains startup information such as:
 The version (including service pack) of SQL Server and Windows;
 Windows process ID;
 Authentication Mode;
 Number of Processors;
 Instance Configuration Parameters;
 Messages for each database that was opened;
 Recovered;
 Successfully started.
 The Error Log also contains informational messages for major events
within the instance, such as traces starting and stopping or database
Backup/Restores.

 The SQL Server error log also contains any messages created with a
RAISERROR command that specifies the WITH LOG parameter.
 Messages are logged to both the SQL Server error log and the
Windows Application Event Log if you execute a RAISERROR with a
severity level of 16 or higher.
 Errors and informational messages related to SQL Server Agent are
found in the SQL Server Agent log file named Sqlagent.out.
 The database mail log is contained in the dbo.sysmail_log table in the
msdb database.
 SSMS has a Log File Viewer that allows you to view the various error
and event log.
 Errors in the master, tempdb and mssqlsystemresource databases an
prevent an instance from starting.

Right
Click
 Log File Viewer

 Database Space Issues
 The most common errors that you encounter deal with running out of space
for either the data or log files.
 When you run out of space in the transaction log :
 All write activity to the database stops.
 You can still read data from the database.
 Returns error 9002 to the application.
 If a transaction log fills up, you can perform the following actions :
 Back up the transaction log.
 Add disk space to the volume that the transaction log file is on.
 Move the log to a volume with more space.
 Increase the size of the transaction log file.
 Add another log file on a disk volume that has space (most common
way).

 Database Space Issues (continue…)
 Transaction log backup might not free up enough space within the log to be
reused if any of the following occurs :
 The database is participating in replication and the distribution
database is not available.
 You have an open transaction.
 Database Mirroring is paused.
 The mirror database is behind the principal.
 A log scan is running.
 SQL Server cannot increase the size of a file quickly enough.
 * Before adding files to filegroups, you should first check if the data
files have the auto-grow feature disabled.
* If auto-grow is disabled and the disk volume still has space, you can
increase the space available just by increasing the size of the existing
files.
* You receive 1101 or 1105 error when the data files are out of space.

 Running out of space in tempdb causes serious problems for a variety of
processes.
 Tempdb storage is used for the following:
 Work tables for GROUP BY, ORDER BY, and UNION queries.
 Work tables for cursors and spool operations.
 Work tables for creating/rebuilding indexes that specify the
SORT_IN_TEMPDB option.
 Work files for hash operations.
 Tempdb storage is also used for the version store.
The version store is used to store row version for the following:
 Online index creation.
 Online index rebuild.
 Transactions running under snapshot isolation level.
 Transactions running against a database with the read committed
snapshot property enabled .
 Multiple Active Result Sets (MARS).
 Database Space Issues (continue…)

 Diagnosing Service Failures
 Finding Service Startup Failures
You can use two utilities to view the state and configuration of SQL Server
services:
 The Windows Services Console .
 SQL Server Configuration Manager.
 The start mode for all SQL Server services for a clustered
instances should be set to Manual.
 Service Accounts
 To start SQL Server, the SQL Server service account needs to have
several permissions, including the following :
 Read and Write access to the folder(s) where the data/log files for
system databases are stored.
 Read and Write permissions on SQL Server registry keys.
 Log On As A Service authority.
 Sysadmin authority inside the SQL Server instance.

 Diagnosing Service Failures
 Finding Service Startup Failures
 If the SQL Server service starts and then immediately shuts down:
 Check if any of the following conditions exist for the service
account:
 Account deleted
 Account locked out
 Account disabled
 Password expired

Locks
 Server uses a locking mechanism to maintain data consistency for
multiuser access.
 An internal process called the Lock Manager determines the appropriate lock
to acquire, how long to retain the lock, and arbitrates when processes are
allowed to modify data such that reads are always consistent.
 Locking refers to the ability of the database server to reserve resources such
as rows of data or pages of an index for the use of one particular user at a
time.
 Problems that locking solves:
 Lost updates : happens any time two independent transactions select the same
row in a table and then update it based on the data that they originally selected.
 Uncommitted dependencies : happens when a record is read while it’s still
being updated, but before the updates are final. To avoid the problem of dirty reads
is to lock data while it’s being written, so no one else can read it before the
changes are final.
 Inconsistent analysis: cause by nonrepeatable reads, which happens when
data is being read by one process while the data’s being written by another
process.
 Phantom reads: occur when an application thinks it has a stable set of data,
but other applications are inserting rows into the data.

Locks
 SQL Server has seven different locking modes and three different
lock types.
 Shared (S) : Used for read operations
 Update (U) : Used for updatable resources
 Exclusive (X) : Used for write operations
 Intent: Used to establish lock hierarchy. Subtypes
are intent shared (IS), intent exclusive (IX) and
shared with intent exclusive (SIX).
 Schema: This lock type is used to ensure that the
schema is not modified when another session
uses it
 Bulk update (BU): Used for bulk copying with
TABLOCK hint, it prevents other sessions to
access the table
 Key-Range: Protects a range of rows

 Shared Locks
 A shared lock is acquired for read operations to prevent the data
being read from changing during the read.
 Because read operations cannot introduce data inconsistencies,
you can have multiple shared locks on the same resource at the
same time.
 Exclusive Lock
 Exclusive lock is acquired on a resource that is being modified
and is held until the modification is complete.
 You can have only one exclusive lock on a resource at a time, and
all other processes needing to access the resource must wait until
the exclusive lock has been released.

 Update Lock
 An update lock is a hybrid of a shared and an exclusive lock.
 Although an update lock is acquired for any update, update
locks can be acquired during any action that requires SQL
Server to first locate the piece of data to be modified.
 An update lock starts out by acquiring a shared lock on
resource until it finds the piece of data that needs to be
modified, the shared lock is then changed to an exclusive lock
while the data is being changed.

 Lock Escalation
 The Lock Manager uses distribution statistics, also used by the
Query Optimizer, to determine which type of lock to acquire.
 Because distribution statistics are not always accurate or don’t
always exist, the Lock Manager has a mechanism called lock
escalation that allows a lock to be promoted to another type.
 SQL Server can escalate a row lock to a table lock, or a page
lock to a table lock.

Resource Description
RID A row identifier used to lock a single row within a heap.
KEY A row lock within an index used to protect key ranges in
serializable transactions.
PAGE An 8-kilobyte (KB) page in a database, such as data or
index pages.
EXTENT A contiguous group of eight pages, such as data or index
pages.
HoBT A heap or B-tree. A lock protecting a B-tree (index) or the
heap data pages in a table that does not have a clustered
index.
TABLE The entire table, including all data and indexes.
FILE A database file.
APPLICATION An application-specified resource.
METADATA Metadata locks.
ALLOCATION_UNIT An allocation unit.
DATABASE The entire database.
 The following table shows levels of Locking Granularity


ISOLATION LEVEL DESCRIPTION
READ UNCOMMITTED Data can be read that has not been committed. Although
an exclusive lock still blocks another exclusive lock, any
read operations ignore an exclusive lock.
READ COMMITTED This is the default isolation level for SQL Server. An
exclusive lock blocks both shared as well as exclusive
locks. A shared lock blocks an exclusive lock. Shared locks
are released as soon as the data has been read.
REPEATABLE READ Exclusive locks block both shared and exclusive locks.
Shared locks block exclusive locks. Shared locks are held
for the duration of the transaction.
READ SERIALIZABLE All the restrictions as the REPEATABLE READ isolation
level. In addition, you cannot insert a new row within the
keyset range currently locked by the transaction. Locks are
held for the duration of the transaction.
SNAPSHOT Uses the row versioning feature to keep shared and
exclusive locks from blocking each other while
maintaining data consistency. A read operation retrieves
data from the version of the row prior to the start of a data
modification operation

 Blocked Processes
 Blocking is the term used when a situation occurs that produces
competing locks on the same resource.
 The second process is blocked from acquiring the lock until the first
resource releases the competing lock.
 A process that is blocked stops executing until the necessary locks
can be acquired.
 Each process that executes a command needs to acquire a lock.
 The locks being requested are queued up by the Lock Manager in
the order that the request is made.
 Although blocking is a normal occurrence within any database that
allows data manipulation by multiple users, you have a problem if
the blocking is severe or lasts for a long time.
 Anyone living in a large metropolitan area has first-hand experience
with blocking.
 When a deadlock is detected, a 1205 error message is returned to
the client and the deadlock is recorded in the SQL Server error log.

 DeadLocks
 When two processes block each other in such a way that
neither process can be resolved, the situation is called a
deadlock.
 When this occurs, the Lock Manager detects the deadlock and
chooses one of the processes to be killed automatically.
 When a deadlock is detected, a 1205 error message is
returned to the client and the deadlock is recorded in the SQL
Server error log.
 In addition, you can use Profiler to capture a deadlock trace,
which allows you to inspect the cause of the deadlock
graphically.
 Deadlocks are transient situations basically caused by bad
timing. If the queries that the two sessions were running
completed in less time, the deadlock might have been avoided.

 Server Side Tracing
 Server-side tracing
 Is the process of having your SQL Server machine save events to a
physical file on that machine without using the profiler client tool.
 An event can be any one of more than 100 different types of events,
including
 SQL:BatchCompleted,
 Login,
 SQL:Stmt Completed,
 Object:Created, and so on.
 Server-side tracing is enabled and controlled by using SQL Server
system-supplied stored procedures and functions.
 With these system-supplied processes, you can identify what to
trace, when to start and stop tracing , what traces are running , and
view trace information stored in the trace file.

 Why Using Server-side Tracing is More Efficient than Using Profiler?
 When performing server-side tracing, SQL Server does not need to
send information back to a client tool.
 Sending information to a client tool causes extra overhead and
increases network traffic.
 Instead of sending information to a client tool, SQL Server stores
the trace information in a physical file on the server.
 This file is written sequentially much like SQL Server writes to the
transaction log.

 Creating a Server-side Trace
 You need :
 To define a trace,
 Add the events you want to collect to the trace,
 Start the trace.
 After doing these tasks, your server-side trace will be running and
gathering information about the events you requested to a trace
file.

 Trace description
 We could use this trace to gather the resources used for each completed
SQL batch, so we are collecting the :
• TextData,
• DatabaseId,
• ApplicationName,
• LoginName,
• StartTime,
• Reads,
• Writes,
• CPU information.
 We use this trace information to identify the resources consumed by each
database or login for a given time period.
 Also, this code allows me to see the T-SQL commands executed in each
batch.

 Starting, and stopping the trace
 To start:
exec sp_trace_setstatus @TraceID, 1
 To stop:
exec sp_trace_setstatus <traceid>, 0
 To close and to remove trace from the server:
exec sp_trace_setstatus <traceid>, 2

 Reviewing the Server-side Trace Information
 To return the trace information in Query Analyzer, you can use a
SELECT statement that calls the function
 Select * from
::fn_trace_gettable('C:testrace_file1.trc',
default)

Review
1. What does the System:Processor Queue Length counter measure?
A. The number of system requests waiting for a processor
B. The number of SQL Server requests waiting for a processor
C. The number of processors actively performing work
D. The amount of time that a processor is in use
2. You are trying to troubleshoot a performance issue at Fabrikam. At about 15
minutes past the hour, on a recurring basis, query performance declines for
about 1 minute before application performance returns to normal. What
tools can you use to diagnose the cause of the performance problems?
(Choose all that apply.)
A. System Monitor
B. Database Engine Tuning Advisor
C. Resource Governor
D. Profiler

3. Which types of SQL Server events are logged to the Windows Application
Event log? (Choose all that apply.)
A. Stack dumps
B. Startup configuration messages
C. Job failures
D. Killed processes
4. You are the DBA at Blue Yonder Airlines and the phone rings. The main ticket
booking application has just gone off-line and cannot be reconnected to the
database. You attempt to connect to the SQL Server and find that it is
unreachable. You find that the service has stopped, and upon inspecting the
error logs, you find a large number of device activation errors. What is the
most likely cause of the problem?
A. Someone deleted the ticketing database files.
B. The disk storage system underneath the master or tempdb databases
went off-line.
C. The disk storage system underneath the ticket booking database went off-
line.
D. The SQL Server service account was locked out
Review (continue)

5. Humongous Insurance has hired you to evaluate its SQL Server
infrastructure, make recommendations, and manage projects to improve the
production environment. During your evaluation, you have learned that
database files are stored across dozens of different drives and mount
points. Data and log files are mixed with backup files. System databases
exist on the same disk drives as user databases. Your first project is to
move all the system databases to separate drives from the user databases.
You will also be moving the tempdb database to dedicated storage on each
instance because many very poorly written queries move massive
quantities of data through tempdb. The on-call DBA is performing the
maintenance and has tested the procedures several times in a lab
environment. You receive a call that the first instance, following the
database moves, does not start. You have verified service account
permissions for the folder containing the master database files, that the
master database files are in the correct location, and the startup parameters
are correct. What is the most likely cause of the problem?
A. The master database is corrupted.
B. The mssql system resource database is corrupted.
C. The service account does not have permissions to the folder containing the tempdb
database files.
D. You have a bad memory module in the server

6. Which of the following are used to locate blocked processes? (Choose all that
apply.)
A. sys.dm_exec_sessions view
B. sys.dm_exec_requests view
C. sys.dm_os_waiting_tasks view
D. sp_who2 system stored procedure

Answers
1. Correct Answer: A
A. Correct: The System:Processor Queue Length indicates the number
of processes at a machine level that are waiting for a processor to be
allocated.
B. Incorrect: The System:Processor Queue Length includes any SQL
Server requests that are waiting for processor resources to be
allocated. The counter also includes requests from any other
applications and the operating system that are waiting on processor
resources.
C. Incorrect: you can derive the number of processors actively performing
work by using the Processor:% Processor Time counter with all
associated instances.
D. Incorrect: The amount of time a given processor is in use is retrieved
from an instance of the Processor:% Processor Time counter.

Answers (continue)
2. Correct Answers: A and D
A. Correct: System Monitor allows you to capture performance counters for
the state of the hardware, operating system, and various SQL Server
subsystems, which can be correlated to a Profiler trace to diagnose a
performance issue.
B. Incorrect: Database Engine Tuning Advisor is used to suggest indexes
and partitions that can optimize the performance of a query.
C. Incorrect: Resource Governor allows you to limit the resources available
to one or more connections.
D. Correct: Profiler allows you to capture the query activity on the instance
and then correlate the queries with performance counters captured using
System Monitor

3. Correct Answers: A and D
A. Correct: A stack dump is a critical error and would be logged to both
the SQL Server error log and the Windows Application Event log.
B. Incorrect: Startup configuration messages are logged only to the SQL
Server error log.
C. Incorrect: Job failures are logged to the SQL Server Agent log.
D. Correct: A killed process is a major event and would be logged to
both the SQL Server error log and the Windows Application Event Log
Answers (continue)

4. Correct Answer: B
A. Incorrect: When a database is online and has a connection, SQL
Server has the files underneath the database open. You cannot
delete a file that is open.
B. Correct: If a disk storage system suddenly goes offline, any
databases with files on the offline storage log device activation
errors. If the device activation errors occur for the master, tempdb,
or mssqlsystemresourcedatabase, the entire instance shuts down.
C. Incorrect: A device activation error for files underneath the ticket
booking database takes the booking database offline, but it does
not cause the entire instance to shut down.
D. Incorrect: If the service account was locked out, any subsequent
attempts to use the account would be refused by Windows.
However, any applications currently running under the account
remain running. So a service account lockout would prevent the
SQL Server from starting, but it does not cause a SQL Server to go
offline or throw device activation errors.
Answers (continue)

5. Correct Answer: C
A. Incorrect: It is possible that the master database is corrupted, but the most likely
cause is one of the changes that you made. Because the master database files
appear to have been moved properly, the permissions are correct, and the startup
parameters are changed correctly, it is not very likely that the master database
files have become corrupted simply by moving them to another location.
B. Incorrect: It is possible that the mssql system resource database is corrupted, but
the most likely cause is due to one of the changes that you made. It is not very
likely that the mssql system resource database files have become corrupted by
moving them to another location.
C. Correct: One of the changes that you made was to move the tempdb database to
a new location. The first step in the move process is to alter the tempdb database
and change the location of the data and log files. After shutting down the instance,
you need to move the files to the new location, before starting the instance
backup. The SQL Server service account also needs Read and Write
permissions on the folder for the tempdb database files. You should check to see
if you are getting any “File not found” errors for the tempdb database. It is much
more likely that SQL Server either cannot find or cannot access the folder for the
tempdb database because this is one of the configuration changes that was made.
D. Incorrect: A bad memory module in the server might cause the instance from
starting up, although it is not very likely
Answers (continue)

6. Correct Answers: B and D
A. Incorrect: The sys.dm_exec_sessionsview gives information about each
connection to the instance but does not contain information about any
actively running requests.
B. Correct: The blocking_session_id column lists the SPID that the
connection is being blocked by.
C. Incorrect: Although the sys.dm_os_waiting_tasksview lists processes that
are waiting on a resource to become available, you cannot locate blocked
processes using this view.
D. Correct: The sp_who2 system stored procedure, which has been part of
the product since the release of SQL Server 6.5, displays the SPID that is
blocking a given process.
Answers (continue)

Sql server lesson12

More Related Content

What's hot (11)

Similar to Sql server lesson12 (20)

More from Ala Qunaibi (12)

Recently uploaded (20)

Sql server lesson12