Crash Course: Data Masking in SQL Server

In SQL Server there are a number of different features that can be used to help meet data protection policies and requirements within an organisation. Usually features such as Transparent Data Encryption (TDE) are widely employed to secure data but this doesn’t necessarily protect against accidental disclosure from persons and applications that might have a user within the database. One such feature that I’ve recently been working with is Dynamic Data Masking or DDM.

What is Dynamic Data Masking (DDM) and Why Should I Use It?

Dynamic Data Masking (DDM) was introduced in SQL Server 2016 (also in Azure SQL Database) and is a method to mask sensitive data from disclosure to non-privileged users. A masking function is applied to the data at column level using a DDL (Data Definition Language) statement.

This can be useful to allow a user the ability to select the row from the table in order to be able to design a query or report but then hand the responsibility for running the object back to a privileged user. This way the user without the necessary authorisation within the organisation never sees the sensitive data by accident. Think contractors and consultants.

DDM is applied to the table schema per column. In this example I apply to an existing AdventureWorks2022 test database using the email function.

USE [AdventureWorks2022]
ALTER TABLE [Person].[EmailAddress]
ALTER COLUMN [EmailAddress] ADD MASKED WITH (FUNCTION = 'email()')

Under normal circumstances a user would be able to select out the table and see the complete email address:

As a user without the UNMASK privilege they will see:

In order for a user in a database to be able to select the unmasked data from the table they must be granted the UNMASK privilege.

There are other types of function available to use and even the ability to implement custom masks as per requirements.

Default – full data masking.
Email – Masks the email after the first letter up to the @ and then the domain up to the suffix.
Random – can be used on numeric types to mask the original value with a random value.
Custom String – creates a custom masking method which exposes the first and last letter only with a padding string in the middle.
Datetime (SQL Server 2022 and above) – allows masking segments of a date and/or time.

There are some technical limitations to DDM such as not being able to apply masks to encrypted columns or computed columns. It also doesn’t fully secure data against malicious users who may find other way around the mask.

Conclusion

Dynamic Data Masking or DDM is just one of a number of features available in Microsoft SQL Server that can be used to protect data from leak or exposure. As SQL Server consultants Digital Incite and Matter Ltd can assist with implementing such features. For assistance with DDM or anything SQL Server related get in touch with us today for an initial consultation.

Finding Your Environmentally Sound Windows 10 End of Support Strategy

Last week I attended a discussion in Leeds hosted by Optimo on the subject “Sustainable Bytes: Green Machines”. Of particular interest to the discussion was the upcoming end of support for Windows 10. What this exactly means is that the OS will continue to function but will no longer receive security updates. Without patching against vulnerabilities those Windows 10 PCs will require removal from an organisation as they will pose an unacceptable risk to security. I have seen and heard various statistics but it’s estimated that between 200 – 600 million PCs won’t be able to support the upgrade and as a result will get disposed of. Somehow.

Old computers. Photo by Daniel Dan via Pexels.

Globally e-waste is an increasingly problematic issue that needs desperate attention. Although many of us will go to the effort to recycle devices at the end of their useful life to us it’s not always the case that IT waste is recycled completely. Not all components can be “economically” separated from each other, some components will contain toxic materials and sometimes e-waste is shipped to faraway places where it’s just dumped anyway. That’s regrettably just the start of the problems too.

As a company Digital Incite and Matter Ltd are committed to finding solutions that address the mountain of e-waste humans are leaving behind. So as there’s one year left until Windows 10 is no longer supported by Microsoft here are some strategies that can be used to prevent your otherwise working devices from ending up as scrap.

The first thing you should do is check the upgrade report in the settings app to see what the exact reason you can’t upgrade to Windows 11 for. It may be as simple and straightforward as the TPM hardware is disabled in the firmware settings. In which case check with the PCs documentation for assistance getting to the firmware settings and which setting to enable. You should then check Microsoft’s list of supported Intel or AMD CPUs for Windows 11 to confirm your device is supported.

You can install Windows 11 on unsupported hardware by various means but we’re assuming that your organisation requires official support without complications either from the device manufacturer or Microsoft. We’ll leave you with the idea surrounding that though.

Purchase Extended Security Upgrades (ESU)

An immediate solution without having to do much is to invest in Extended Security Upgrades. The cost of which is $61 per device for year 1. The cost increases each year but there is a discount of 25% if you use Microsoft’s cloud management tools such as Intune. Anyone using Windows 365 receives this for free as well and we understand that educational establishments will also be given a significant discount.

The downside to this solution is that it incurs a significant cost especially in organisations that have many Windows 10 devices. You should also double check with your software vendors as to their support policy for their apps. You may find they drop support for Windows 10 anyway. It also needs to be noted that this solution is for security updates only. There are no bug-fixes, design changes or additional technical support given through this.

We recommend this solution sparingly as the investment can be substantial. If your organisation has a handful of devices nearing the end of service (circa 10 years we try to aim for) then this might prove a good choice in the short-term as opposed to purchasing new devices straight away.

Replace with an Alternative OS

Windows is a favoured choice of many organisations but it’s not the only OS out there. Linux distributions like Ubuntu are increasingly popular and viable in the workplace. Such operating systems are not just trusted, reliable and secure but are also free of cost. This makes it an economical solution for replacing Windows.

A potential downside to using a Linux distro is the software support. Not all application vendors offer support for Linux. Microsoft Office for example does not have support for Linux. Alternatives such as LibreOffice or Collabora office are widely but not completely compatible with Microsoft Office document formats and as such may not 100% fit in with your workflow.

Replacing Windows 10 with a Linux distribution is a great solution as it will allow continued use of the hardware. Organisations that use a browser based workflow should strongly consider this

Reworks

If the only reason you can’t get Windows 11 installed is hardware limitations and alternative OSes aren’t possible then another avenue to explore is the possibility of upgrading or replacing the internals of the device to address whatever incompatibility that has arisen.

If the issue is a lack of support for a TPM (Trusted Platform Module) then the chip can usually be attached to the motherboard header of a desktop PC and can be acquired for around £20. Do check to see if your device just needs firmware based TPM enabled in the settings first however.

Alternatively if the CPU is not supported then looking at upgrading hardware is a nearly last resort. It should be noted that Windows licences are tied to the motherboard (the licence is stored as a firmware variable). As there would be almost certain to have to upgrade the system motherboard to gain a supported CPU (particularly Intel CPUs which generally change socket every generation) you may find having to acquire another licence for Windows anyway. Despite this potentially chassis, power supply, memory, and storage devices can be reused. This would effectively make your device a “custom build” which may still be a good option for higher end devices. Otherwise you may find some of the parts – particularly storage devices, GPUs, and sometimes memory – as useful spares or upgrades for other PCs.

This option is unfortunately a little bit trickier than the others given often limited options particularly around finding compatible upgrades but it’s still worth consideration.

Conclusion

Although the end of Windows 10 means that some hardware becomes “obsolete” it’s definitely not the end of the line.

We call upon Microsoft and their hardware partners to support devices for as long as possible instead of planned obsolescence. The needs of the planet and human society need consideration when an OS becomes “unsupported” rather than shareholder first marketing plans to sell new hardware when the ecosystem struggles to deal with the disposal of the hardware already in circulation. Some models of laptop and unfortunately PCs are becoming harder to upgrade & repair and we don’t agree with that principle at all.

Digital Incite and Matter Ltd are here to help when it comes to creating an environmentally considerate strategy around hardware procurement, service and end of life policy for your organisation. Feel free to get in touch with us for help and assistance.

Microsoft SQL Server 2025 Announced

The next version of SQL Server has been announced at Microsoft IGNITE 2024: SQL Server 2025 will be the next release for the long-running RDMBS. vNext is currently in private preview and will have a heavy focus on AI capabilities. For example vector data used for machine learning is supported as well as support for direct REST interfaces for AI services.

Other particular highlights we’re anticipating include:

Support for Entra ID Managed Identities.
Enhanced performance in the database engine including lock optimisations.
Support for JSON as a native data format.
Support for Regular Expressions (REGEXP_LIKE) in T-SQL.

In addition to the SQL Server platform itself Microsoft also announced that SQL Server Management Studio (SSMS) will receive a number of updates including Git integration, Copilot AI support for help with writing code as well as a much requested dark mode in the UI.

It’s pleasing to see a new release of SQL Server that delivers notable features already introduced in Azure SQL to a product that customers may use in their data centre or via IaaS in the cloud. This will allow customers to utilise a supported release of SQL Server well into 2035 giving enhanced choice in a fluid computing market.

We’re expecting the final release of SQL Server 2025 sometime next year following a private and then public preview. Like all SQL Server releases Digital Incite and Matter Ltd will be at the forefront of new developments and look forward to working with you for your data platform needs. Get in touch with us today to discuss your requirements.

Tips for Deploying Secure Instances of Microsoft SQL Server

Your organisation’s data is a critical asset that needs the upmost care and attention to securely store & manage. For organisations with a Microsoft technology stack – in terms of infrastructure, applications and internal skills – Microsoft SQL Server is a RDBMS of choice designed with security in mind to meet such requirements.

When you deploy an SQL Server instance it’s important to work with your internal IT teams to make your installation secure right from the point of installation. In this blog post I’m going to discuss 5 points I recommend to ensure a secure basis for any SQL Server installation.

Tip 1: Deploy on Windows Server Core. If you can.

Most of us are familiar with the Windows GUI but for Windows Server it’s possible to deploy a reduced install of the OS known as Windows Server Core. This feature was first introduced with Windows Server 2008. This is achieved via a selection of the edition when the OS is first installed and is actually the default choice.

Installing Windows Server as Server Core affords a number of advantages. The principle benefit is that as the install is reduced in terms of components and features compared to a GUI install the attack surface is much reduced. There is also a slight performance advantage in terms of CPU and memory saved by the reduced overhead of the install.

Note that a drawback of a Server Core install is that some features such as SQL Server Reporting Services need the GUI install. If you choose to deploy an application on the same server (typically to save costs if the application only serves a few users) the vendor may also require a full GUI install. You also need to be completely proficient in using CLI based tools or remote management as the usual GUI tools aren’t present.

Tip 2: Apply the Latest Cumulative Update or Security Update

SQL Server generally has a good record when it comes to security vulnerabilities. That being said there has been a number of security issues identified and fixed in SQL Server. Patching just like any other program in use is a necessity.

In SQL Server you have two options regarding your update track. Highly recommended is the apply the latest Cumulative Update (CU). Not only will the instance receive the most up-to-date security fixes but also fixes to other areas of the database server.

Alternatively if your SQL Server instance hasn’t been patched using a CU then it would still be considered “GDR” or General Distribution Release”. These releases will still receive updates without applying a CU typically via Windows Update.

You should also check that on both clients and server that the various SQL Server connectivity drivers are patched up to date if they are mandated by any client applications.

Tip 2: Enable TLS by Applying a Certificate

TLS or Transport Layer Security is an encryption protocol that protects data sent and received between SQL Server and the client. Under principles of Zero Trust you should still take steps to encrypt data in transit even if you own the network SQL Server is running on.

A certificate should be procured either via an internal Certificate Authority (CA) server or via a recognised broker. For testing purposes a self-signed certificate can be used by is not recommended. This then secures application traffic against eavesdropping.

Note that you should never allow connectivity to SQL Server over the internet.

Tip 4: Use Windows Authentication Where Possible

SQL Server can be run in either Windows or Mixed authentication modes. There are a number of drawbacks to using SQL Server’s built-in authentication method. Chiefly is that login passwords get transmitted unencrypted and can be seen using packet sniffing.

For best results leave use Windows Authentication mode only and make sure your applications authenticate via Windows. Certain applications do not support Windows authentication so check carefully with the vendor in case any patches or reconfiguration are required.

Tip 4a: If You Must Use Mixed Authentication Disable the ‘sa’ Account.

The sa account in SQL Server is a very special one. It has the SID of 0x01 and is considered the most powerful account in SQL Server. It’s also the most vulnerable one being so well known. For best results leave sa disabled and rename it. That way it can’t be used to exploit the whole instance.

Tip 5: Consider Firewall Rules

The Windows or Linux firewall should be enabled on each operating system for SQL Server. Connectivity to SQL Server is via the default port 1433/tcp. A rule needs to be in place to allow traffic to and from the server.

Locking down firewall rules to only allow access via known application servers, authorised management PCs and restricted virtual desktops is highly recommended. This effectively stops any malicious login attempts from unauthorised hosts. This can help keep your SQL Server instance safe.

Conclusion

In this post we’ve discussed the importance of protecting your organisation’s data as well as the theory over a number of basic steps for securely deploying SQL Server instances including: secure configuration, updates, firewall controls as well as encryption in transit.

Once the SQL Server instance is promoted to production there are also other features of SQL Server that should be considered such Transparent Data Encryption (TDE) for encrypting SQL Server databases, logs and backups.

If your organisation is facing challenges over securing your SQL Server instances then help is at hand. Digital Incite and Matter Ltd can help further with this challenging subject. We also have experience helping clients with annual security audits. Get in touch with us today for a free initial consultation.

Windows Server 2025 Now Available

Microsoft have recently announced the availability of the next version of Windows Server. Through retail or channel partners Windows Server 2025 is now ready for deployment in both your own datacentre or in public cloud environments.

There are many areas of new features and improvements in Windows Server 2025 including enhancements to Active Directory Domain Services (ADDS), security, performance, software defined networking, general management, and much more.

As with all previous releases of Windows – server and desktop – it will take some time before applications are certified to run on the new release. That being said it’s well worth evaluating Windows Server 2025 for any near term deployments you may be considering. Although Windows Server 2022 will continue to be generally supported moving towards the latest version of the Microsoft server system is recommended for maximum longevity and taking advantage of the latest improvements.

Whether you’re considering a new deployment, a cloud repatriation strategy to reduce costs or need to move away from older versions of Windows Server that are no longer supported Digital Incite and Matter Ltd have the knowledge and expertise to help. Please contact us today to discuss your upcoming projects.

Something Went Wrong – A Crash Course in Providing a SQL Trace File

We’re all (probably) familiar with the error message Something Went Wrong. You might also receive a long string of numbers, get told to try again later and report the error to an administrator. These kind of messages aren’t really helpful by themselves.

Regardless of where this error is coming from you need to get a better understanding of what’s actually going wrong in order to be able to work out what to do to address it or even to determine if you need to do anything about it.

In SQL Server there are two features which can be used to provide in-depth information: the SQL Server Profiler and Extended Events. As a SQL Server and/or application administrator learning to use these tools is a critical skill in responding to application errors.

Before we go into any further depth it’s necessary to mention that SQL Server Profiler and also the SQL Trace events are marked as deprecated by Microsoft. This means that the features are no longer being developed and will be removed in future versions of SQL Server. That also means you should learn Extended Events for future reference.

From professional experience however many application vendors we work with will still request an SQL Server Trace file (with a .trc extension) so if you need such a trace to add to a support ticket read on.

A Basic SQL Server Trace

Before any trace is started on the SQL Server instance please be aware that this activity can…will place a significant strain on server resources. This could either be the amount of disk space required for the capture or the impact to processing performance. With this in mind you should try to organise an SQL Trace session when you have either exclusive access to the application database or that you know that it will be a quiet time (i.e. after regular office hours). If you are doing this for the first time definitely practice on a test/training environment before working on production.

The settings I’ll go through here are to be considered a basic trace only. If you are being asked to provide a trace file you should always first check with the software author or your DBA asking what events they need to see in the trace. You can then either open their trace template or use the instructions below to select the trace criteria they require.

Please note that the following was performed on SQL Server 2022 with SQL Server Management Studio (SSMS) 20.2. The SQL Server Profiler has very much been the same with all prior versions so don’t expect any major deviations in the instructions.

Step 1: Launch SQL Server Profiler from either the start menu or from SQL Server Management Studio from Tools > SQL Server Profiler. Login with a user that has permissions for ALTER TRACE by clicking Connect.

Step 2: On the first tab – General – we need to setup the basics of the trace:

Give the trace a suitable name.
For a template use Standard (default) for now.
Check Save to File. This will then prompt a save location.
- Set maximum file size (MB): up this to 100 MB
- I recommend leaving Enable file rollover checked. This will create multiple trace files as necessary.

Step 3: On the second tab – Events Selection – you will now select the events you want to capture. As we selected the Standard (default) template we have a set of pre-selected events to work with already. I suggest adding a few more in order to get a slightly more useful trace:

Click Show all events and Show all columns to see the full list.
Select the additional rows using the checkbox to the left of the event:
- Errors and Warnings
  - User Error Message
- Stored Procedures
  - SP:Completed
  - SP:Starting
Finally uncheck Show all events to only display what events you have selected. Have a little review to make sure the ones suggested have been selected.

Ideally we should filter the trace events to those for the required database(s) only. This can be done by clicking the Column Filters button. On the left scroll to find the DatabaseName filter. Expand the Like operator on the right of the Window then type in the name of the database you want to Trace for. Click OK once you have your databases listed.

Step 4: Click run to start the trace. You will see the trace window appear with events being logged. Don’t worry if the events are flying past too quick. The SQL Server Profiler by default will continuously scroll to the bottom.

Notice that there is also now a trace file saved to the path you specified in Step 2.

Step 5: Either get your colleague to replicate the problem in the application or follow their replication steps yourself. In this example I captured doing a simple select statement from AdventureWorks.

At this point note the two highlighted buttons on the screenshot below. The left “stop” button (Stop Selected Trace) will end the trace and stop SQL Server Profiler from capturing events. The right button (Auto Scroll Window) will stop the window scrolling if you need to quickly study a series of events you have noticed whilst keeping the trace capturing new events.

Once you have finished live analysis make sure to click the Stop Selected Trace button. As mentioned SQL Server Profiler running a trace will have a significant performance impact to the SQL Server instance so don’t leave it!

The aforementioned trace file can now be securely transferred to the software author or you can re-open it later and review it any time you want.

In the SQL Server Profiler window you’ll no doubt see many different events and also captured text in the bottom half of the Profiler window. By studying what the SQL Server engine is doing we can begin the process of troubleshooting problematic or unexpected application behaviour. Whilst the level of information in an SQL Server trace will be comprehensive it’s necessary to take the time to study it properly.

Conclusion

In this blog post we’ve learned to create a basic trace file in SQL Server. Whilst only a basic trace additional events can be captured as well as additional filters specified to help us understand application behaviour when an issue is reported.

If you need further support Digital Incite and Matter can not only help create the requested trace file but we can also work with your software provider to manage the incident case from diagnosis to patch deployment. Please get in touch with us today for further assistance.

Why You Should Repair and Not Replace Your Organisation’s Devices

IT related waste is a growing problem worldwide. In 2022 alone over 62 million tons of “e-waste” was generated. Whilst there has been a number of initiatives worldwide to alleviate this problem such as the EU directives for common power chargers for portable devices and WEEE for device recycling the amount of global e-waste continues to grow rapidly.

Device support policies are a significant cause of e-waste. On October 14th 2025 Microsoft will end support for Windows 10 Home and Pro. Up to and past that important date IT teams across the planet will be replacing devices that will no-longer receive important security updates. This will sadly trigger otherwise worthy devices to be sent for recycling or even landfill if not handled properly.

Not only do important OS end of support dates trigger unnecessary device replacements but also hardware designs that make it difficult if not impossible for a user to replace faulty parts which further contributes to the e-waste problem. For example use of adhesives in bonding together tablets and smartphones can make it a challenge to open a device for repair without shattering a screen.

Thankfully it’s not all bad news. The right to repair movement is still strong and gaining further momentum. Many devices out there – desktops, laptops and servers – are still designed with upgradability and repair in mind. For the impending Windows 10 end of support date the Operating System can be changed for a Linux OS allowing a device to be continued to use.

When choosing devices we always encourage sourcing from manufacturer’s that have serviceability considered in their product design. This also includes open warranties, spare parts availability and avoiding Operating System lockdown.

Digital Incite and Matter Ltd can provide a variety of servicing, upgrade and repairs for your desktop and laptops. Whether you require an upgrade to storage, reload of the operating system or a device repair we can be of assistance. Not only does this save on replacing devices that would otherwise only need an economical repair it also helps your organisation meet environmental goals by avoiding sending devices to landfill.

For more information please get in touch with us.

Supporting World Mental Health Day

Masks off. This week on October 10th 2024 is World Mental Health day. This year’s theme is very on point for professionals worldwide: “It is time to Prioritize Mental Health in the Workplace”.

According to the charity Mind statistically around 1 in 4 people in England will experience a mental health issue each year with specific demographics such as LGBTQIA+, Black or Black British people, Young Women and those with overlapping problems even more likely to report mental health problems.

Thankfully the conversation is moving on and the stigma around having mental health problems is being challenged. In a work setting it’s of critical importance to create a mental well-being supportive environment. Whether it’s additional support from colleagues, time off to seek help and treatment or reworking a job role everyone should be able to get help when they need it most.

I volunteer for Andy’s Man Club and they’re a great example of a group that’s worked hard to challenge the stigma and support men like me when life got challenging. For other groups that cater to your needs check out Hub of Hope for links to other groups.

Multi-Factor Matters

Did you know that 25% of users in a World Password Day survey admitted to reusing their passwords across multiple sites? This kind of behaviour poses a challenge for organisations. Should a user’s password be guessed or compromised then an attacker could access multiple systems via that one combination.

Multi-Factor Authentication or MFA is a system by which an additional factor of authentication beyond a password is added to an account. This is done to enhance the security of the account by preventing takeovers if the password is lost, guessed or brute-forced. You may also hear this referred to as Two-Factor Authentication or 2FA which is a term often used interchangeably.

By using MFA for both administrators and users of a system you can prevent account takeovers that result from passwords that have been guessed, reused or compromised. Enabling MFA for any cloud services your organisation subscribes to is also required for schemes like Cyber Essentials in the UK.

Types of MFA could include but are definitely not limited to:

A one-time code sent via: SMS to a mobile phone number, voice to a telephone number or an email account
Response to a notification via an authentication app on a smartphone.
a Time-based One Time Password (TOTP) generated by a smartphone app or physical device.
A hardware authentication device such as a Yubikey. This may support the FIDO2 scheme and/or the older U2F standard.

Yubikey atatched to a keyring — An example of MFA: a Yubikey.

It should be noted that whilst MFA offers additional security from unauthorised access it does not completely guarantee secure systems. Breaches may still result via other means. Principally you should always be mindful of attackers trying to gain access to an account by social engineering techniques. A common way by which this is happening is attackers calling users to ask them for the one-time codes.

Additional ways of adding security to accounts should also be considered. A password manager can help users avoid password reuse, help them to generate secure & unique passwords per login and also allow an administrator to monitor password use. Securing sites with Single Sign-On (SSO) is also another option to explore. This allows a user to access multiple systems seamlessly via one login. This can help users by avoiding situations where they struggle to remember different passwords and be tempted to reuse the same password. Instead their “primary” account acts as the login which can be secured via MFA and continuously monitored (such as Microsoft’s Conditional Access for Entra ID).

Ultimately the password is indeed quite dead. It’s not uncommon that in our day to day lives across personal and professional accounts an individual person might have to remember 100s of individual accounts. Passwordless schemes along with SSO are the way to go.

In summary MFA is a great way of adding additional security to a system for both administrators and users. It should be an automatic requirement in setting up new accounts – both cloud and privately hosted – especially in Cyber Essentials certified organisations.

Get in touch with us today for further assistance with Active Directory, Entra ID or SQL Server based authentication challenges.

Performance Tuning SQL Server Queries

Let’s imagine that you’ve just come back from summer holidays. You’ve been away travelling, enjoying life and having a relaxing time. Crash down back to work in September (sadly) and you’ve been asked to write some T-SQL based queries. Maybe it’s for a new dashboard component, report lines or even a new view. Somehow that query isn’t performing as you or a colleague expect. Panic sets in as you look blankly at the query and think “what do I do?!?!”.

Don’t fret; performance issues with T-SQL queries happen. Most of the time it’s an issue that can be fixed very quickly. If you find you’ve gone a bit rusty over summer here’s five quick tips to help you troubleshoot those queries:

Tip 1: Check For Obvious Offenders

Quite often you’ll find your issue by re-reading your T-SQL code and making some adjustments. Common causes could be:

Avoid Using SELECT *. Whilst this is sometimes OK to get an understanding of the table contents what it does result in absolutely everything in the table being retrieved. Not only is the full set of data unnecessary but you may also be impacting other queries executing simultaneously on the instance.
Check for complex joins. Occasionally a table join will introduce a complex operation that SQL Server must complete to get the matching rows. Check that your joins are appropriate for the data you want to select.
Consider using the WITH(NOLOCK) hint on live OLTP databases. This prevents queries from locking the table and blocking other queries. This may result in dirty reads so watch out!
Not obvious where the query is going slow? Break the query down bit by bit and re-introduce statements one at a time until you find the offending part of the query.

Tip 2: Check the Query Execution Plan

An automatic go-to for understanding how SQL Server is working at the engine level is to include the Execution Plan. In SQL Server Management Studio (SSMS) you can enable this option on the menu bar at the top.

SQL Server Management Studio showing how to enable Execution Plan

Once your query has been executed and completed the actual query execution plan will be displayed as a tab next to the results window. Within the execution plan check for expensive operations, missing indexes and also any other pertinent issues that might be causing your problem. Some execution plans end up quite in depth so take some time to study each part of the plan.

Once you have an understanding of you SQL Server is trying to perform your query you can then go make adjustments to your T-SQL or work with your DBA on the potential bottlenecks the execution plan helped you identify.

Tip 3: Check Indexes

In SQL Server Indexes are used to help the database engine perform the most efficient lookup of data possible. Having indexes in place is imperative to a production database especially where the number of records becomes substantial. Without indexes the database engine must perform more complex lookups of the data held in a table which takes longer to complete versus tables that have appropriate indexes designed.

Check the execution plan for Index Scans and Seeks. Index Scans are more intensive as they mean that the whole table is being looked up whereas an Index Seek means only the matching records. An Index Seek is generally preferable except in certain circumstances where there are a large numbers of matching records in a table. You will also see a warning box at the top of the execution plan if SQL Server has identified a missing index that may have helped the query.

Tip 4: Index Defragmentation and Statistics Update Jobs

As discussed in the previous tip indexes and statistics are important in maintaining adequate database query performance. Over time the indexes on a table will become fragmented and the statistics become out of date. This will gradually harm performance over the course of time. If you find your existing queries are getting slower, slower and slower then this is a probable cause.

Check the system view sys.dm_db_index_physical_stats for your particular database. If the avg_fragmentation_in_percent value for your indexes are running high check on the SQL Server instance that an Agent job or maintenance plan is in place to perform a regular index reorganise or an index rebuild for heavily fragmented indexes.

Depending on how the SQL Server instance is configured statistics may be updated automatically however there should also be an SQL Server Agent job or maintenance plan to update index and/or column statistics on a regular basis as appropriate.

Tip 5: Use the Query Store

The Query Store is a useful feature that was introduced in SQL Server 2016. It is not enabled by default unless you are working with SQL Server 2022 and have created a new database. Not only does the Query Store contain a number of useful reports that help you understand how queries are performing in your SQL Server instance but from SQL Server 2017 it enables the Automatic Tuning functionality.

If the Query Store is not enabled then it may be enabled on a per database via SQL Server Management Studio or T-SQL like so:

ALTER DATABASE <my_database>
SET QUERY_STORE = ON (WITH OPERATION_MODE = READ_WRITE)

Once the Query Store is on you need to let it run for a while to let it capture sufficient data. Whilst this is going on take some time to review the Query Store for information on query regressions and potentially make adjustments to the query accordingly.

Once you have captured sufficient data you can then enable automatic tuning on your database like so:

ALTER DATABASE <my_database>
SET AUTOMATIC_TUNING ( FORCE_LAST_GOOD_PLAN = ON );

Still Having Issues?

If you find yourself still having performance issues then perhaps it’s time to bring in a consultant to help you out. Digital Incite and Matter are proficient at query writing and optimisation. If you issue lies beyond the obvious then we can also advise on the infrastructure, instance configuration and expanded troubleshooting.

Get in touch with us today and we’ll be pleased to assist you further.