Taking over a database you’re not familiar with requires you to dedicate some time to analyze and understand it. Deciphering an inherited database takes some time, that among other factors, depends on the available documentation (if any), comments left in object SQL, and database complexity.April 18, 2013
Even though Dynamics CRM 2016 is very on clear on the fact that that record deletion is permanent and even provides a warning, it’s not unusual one or more CRM records to be deleted by accident. The bad news is that once the records have been deleted, there is no way to undo the deletion as removing a record from CRM actually translates to deleting it from the underlying SQL database.
That is one of the biggest differences between Dynamics CRM 4 and Dynamics CRM 2016. Unlike Dynamics CRM 2016, Dynamics CRM 4 uses the “soft delete” approach. This means that a deleted record in Dynamics CRM 4 could be recovered relatively effortlessly – to recover a deleted record from Dynamics CRM 4 all you have to do is to connect to the underlying database, locate the deleted record and simply update the value in that record’s IsDeleted column to false. So, how can one go about recovering deleted records from a Dynamics CRM 2016 database?April 12, 2013
Being able to revert to the latest available backup immediately is a key requirement that every SQL Server DBA needs to meet. There are so many scenarios in which immediate action needs to be taken – starting from actual disk failure or corruption, system failure all the way to UPDATEs without WHERE clauses, malicious INSERTs or accidental DELETEs, DROPs or TRUNCATE operations. However, usually this is easier said than done
Even if you take meticulous care of your database backups, have the last available database backup verified and at hand and are sure that restoring it won’t overwrite any valid changes made to the database since the accident has occurred – the fact that the system will experience downtime until the backup is fully restored still remains. Luckily, downtime caused by the time needed for a full database backup restore can be minimizedApril 12, 2013
The time when source control systems were used only for code development is gone. Today, database versioning is quite common, even in small companies.
As there are many different source control systems, with different features and options, there are no strictly defined rules how to use them. Each company can create its own standards and rules for using a source control system for database versioning, following the best practices and recommendations.
Depending on which source control system is used, and what the usage schema and environment flow are, the goals of using source control systems are safety, easier team-based database development, and deployment automation.
In a multi-environment system, each environment represents a state through which a database must pass, until it reaches its final destination – production. For example, development of new features and fixing of existing issues is carried out in Development, then they are tested in Integration (or Testing) and Production is where your live databases are.
One of the common rules is that all deployments come from source control. In other words, all database changes must go into source control first and then be deployed to production.
Does it always have to be so?
It should, but unfortunately, it isn’t. Despite all the benefits that a source control system brings, the key factor is still human. If there’s no discipline, developers can skip a step or two, ignore a rule, or do something that doesn’t comply with the standards. A developer in a time crunch might have applied a change directly to production, skipping both the development and testing environments, and never adding the change to source control.
The good news is – in a versioned environment, this can be easily fixed. Furthermore, possible problems caused by this undisciplined approach can be prevented by checking the production database before deployment.
No matter how small the change is, and how small the difference between the production and the source control version is – it’s still a difference. Even just changing a column data type from int to e.g. nvarchar(50) in production, makes a difference that can cause headaches when deploying a new version from source control. Next time the new version is deployed to production, either scheduled, or on demand, deployment can fail.
If lucky, all that will happen is that the latest version cannot be deployed. In a worse scenario, even some data can be lost.
How can be made sure there were no direct modifications of your production database?
Before the deployment starts, check whether there are any differences between the production database and the source control system version that was deployed last. Another option is to compare the live database to a full database backup created immediately after the last deployment if they are available.
ApexSQL Diff is a SQL Server database comparison and synchronization tool which detects differences between database objects in live databases, full and differential database backups, databases versioned in source control systems, script folders and database snapshots. It generates comprehensive reports for the differences and can automate the synchronization process.
- Start ApexSQL Diff
Once the New project window is shown set the following in the Data sources tab:
- Select Database as a source type
- Select the production database as the source
- Select Backup as a destination type
- Select the full database backup taken immediately after the last deployment as the destination
- Click the Compare button
To switch how to view differences, on the View tab, click one of the buttons under the Grouping section. By default, the Group by difference type option is selected:
- All objects in the Results grid will be grouped as equal, different, in source or in destination only.
If there are no differences, all objects will be shown as equal.
If there are any different objects, check out their differences in the Script difference view panel:
- To create a synchronization script, check the objects in the Results grid that should be synchronized
Click the Synchronize button from the Home tab to initiate the Synchronization wizard:
- In the first step of the Synchronization wizard, the synchronization direction will be shown, while in the second step, any dependent objects will be automatically checked and added in the synchronization script:
- Under the Output options step, leave the Create a synchronization script action and choose the Save script to a file option:
In the last step, review the Summary and warnings and click the Create script button to finish the process:
Get back to developers, investigate the differences, and decide the best course of action in this specific situation. The generated script can be used in desired sandbox to synchronize the testing database with the production one.
To avoid deployment problems, caused by changes in production that didn’t come from source control, use SQL server comparison tool ApexSQL Diff. If there are any differences detected in production, it’s not recommended to make any further changes to the production database without testing them first.
One of the ways to keep stored procedure, function or view schema hidden is using the WITH ENCRYPTION statements. In this way, access to objects’ DDL script can be locked, so that unauthorized users cannot see them:
CREATE FUNCTION [dbo].[ufnGetAccountingEndDateEnc]() RETURNS [datetime] with encryption AS BEGIN RETURN DATEADD(millisecond, -2, CONVERT(datetime, '20040701', 112)); END; GO
There are certain situations when creating a build script from a database is necessary – when a specific object (table, view, stored procedure, user, etc.) needs to be recreated or rolled back to a state it had earlier. If an online database doesn’t contain this object anymore or contains a newer version of it, the only places to look at are full database backups. If a database is versioned under source control, an object can be easily found there, but let’s assume that only SQL backups are available.
The most logical solution is to take a full database backup, restore it, and get what is needed and in most cases, this is the best and quickest solution.
This is a perfectly viable solution for small databases where backups take several hundred of MBs. For large databases, besides significant time needed to restore a backup, the lack of space on hard drives might be encountered as well. The time needed for a database restore depends not only on the backup size, but also on the server configuration and current load. The space needed is at least equal to the size of the full database backup.
ApexSQL Diff is a SQL Server database comparison and synchronization tool which detects differences between database objects in live databases, backups, snapshots, script folders and source control projects. Although it is most commonly used for comparison and synchronization, its features can be used to achieve some other tasks. Here is a simple trick used for this solution: comparing a data source such as a live database, a database in a source control system, a backup, or a snapshot to a blank destination will actually create a build script that creates SQL objects that exist in the source.
This means that with ApexSQL Diff there is neither the need for additional space on hard drives, nor the need to restore a database backup; at the same time a DDL script can be created for the selected object in the database. If there are any dependent database objects, they can be scripted too and will be created in a proper order.
The process is as following:
- Create a new folder. This is the folder where individual build scripts for each scripted object will be saved
- Start ApexSQL Diff
- Click the New button in the Project management window:
In the Source panel:
- Select Backup from the Source drop-down list
- Click the Add button and navigate to the folder where the backup file resides
- Select the backup and click the Open button
- In the Destination panel:
- For additional object filtering prior to the comparison process, check out article on this link
- Click the Compare button in the bottom-right corner of the New project window
- The objects read from the database backup are shown in the Results grid. As the destination script folder is empty, all objects are shown as missing:
- Check the check-boxes to select the objects that need to be scripted:
On the Home tab, click the Synchronize button:
- In the Synchronization direction step of the Synchronization wizard, click the Next button
The Dependencies step shows the objects that the objects selected for synchronization depend on. By default, they are all selected:
Uncheck the Include dependent database objects checkbox to avoid updating objects other than the ones selected in the Results grid.
In the Output options step, select the Synchronize to script folder as an output action to have a DDL script created in the script folder for each of the selected database objects:
To create a single deployment script that creates all objects selected in the Results grid, select the Create a synchronization script output action:
- In the Summary and warnings step, review the order of the actions that will be made and click the Synchronize or Create script button:
Depending on which output action was chosen, two buttons can be shown in the last step and the following will be done:
The Synchronize button – if the Synchronize to script folder action was chosen, separate sub-folder will be created for each object type has, such as Tables, Functions, etc. The database settings are scripted in the DatabaseSettings.xml file, saved in the specified script folder:
- The Synchronize button – if the Synchronize to script folder action was chosen, separate sub-folder will be created for each object type has, such as Tables, Functions, etc. The database settings are scripted in the DatabaseSettings.xml file, saved in the specified script folder:
The Create script button – if the Create a synchronization script action was chosen, the created script will be created and it can be executed directly from the Integrated editor, with a click on the F5 key. If dependent objects are created, there is no need to worry which object will be created first; ApexSQL Diff determines the correct object creation order and generates the script:
A database backup doesn’t have to be a black box accessible only when it’s restored to a SQL Server. Database comparison and synchronization tool – ApexSQL Diff can read its content without restoring it and thus help out to work with SQL objects stored in it.
Sometimes the best test data is the data you have in your live, production SQL Server database. Since using production data is most often not an acceptable option, this requires retrieving the data from a live database table and inserting it into a new table. The challenge arises if you don’t want to insert all records from the original table but only the records that comply with certain conditions.
How to create INSERT INTO script that moves only specific data from one table to another.April 5, 2013
When work on a database in a development environment is completed, developers are required to send the new version in for testing. If the testing is done on one or two machines – this is not a problem. However, if there is a need to test the new version on 10 or 15 machines (with different operating systems, SQL Server versions, or service packs installed), this can be tedious.
Another situation where scripts need to be executed on multiple servers is if there are many databases for different clients that were all derived from the same database and share many common objects. Any changes made in the development must be pushed to all users.April 4, 2013
Even with all precautions taken, mishaps with databases are still a possibility. Although it is good to be prepared for a disaster – the database is in the full recovery model and transaction log and database backups are created regularly, there are still some situations when a problem cannot be solved using an out-of-the-box solution.
One of these issues is restoring a single object. It’s not unusual to delete a SQL database object (e.g. a stored procedure) that is still needed, or modify it when the old version might be required later.April 4, 2013
There are numerous scenarios when data synchronization between the two databases is needed, such as distributing data to different locations, integration of data from different locations into a central repository, or just a simple synchronization between a test and production database.
The recommended solution is to use database replication.
However, even with database replication there’s a need to check whether all the data is synchronized and to synchronize it if needed.April 4, 2013
It’s common knowledge that running database changes directly in production, without thorough testing first, should be avoided.
If there are enough resources in the environment, there would be at least one testing and one production SQL Server instance. However, that introduces another challenge. When everything is set up in the test, and runs smoothly and as expected, how can it easily be applied to the production instance?April 4, 2013
Collaborating on database development introduces a series of challenges. For instance, having the development team change the tables, views, stored procedures and other objects in a single, shared database, although intuitive, can introduce severe issues down the road as valid changes can be lost or overwritten by an unsuspecting teammate. This issue might be mitigated by restoring a database backup – under the assumption a valid backup exists. Even if it does, overwriting the development database with a backup means losing all of the valid database changes that have occurred since the backup was taken; not to mention the fact that restoring a large backup can take time – during which none of the developers can work on the database being restored.April 4, 2013
In scenarios where a development project of a large scale is being completed, DBA skills can really be put to the test. Due to the nature and the dependencies of the software which is being developed, it’s not uncommon for the DBA to be in charge of dozens of environments, used by multiple development teams working on different aspects of the project. In some cases, this scale of development doesn’t only require for multiple SQL Server instances to be set up across environments – more often than not it requires different versions of SQL Server to be accessible across environments as wellApril 4, 2013
Frequently, developers and DBAs need to search databases for objects or data. If a database function that contains a specific table column or a variable name, or for a table that contains specific data is searched, the simple solution for finding those values, does not exists.April 4, 2013
The first part of the series – SQL Formatting standards – Capitalization, Indentation, Comments, Parenthesis, explains the importance of having clean SQL code. In short, deciphering someone else’s code is time-consuming. Clean and neat SQL code can be read faster; SQL reviewing and troubleshooting is more efficient; joint development efforts are more effective; handing off projects from one team to another is smoother than for inconsistently written SQL.
As there are neither style nor standards to format SQL, it’s up to the team to create its own set of formatting standards. Here are some recommendations to format joins, value lists, code structure, arithmetic, comparison and logical operations.April 4, 2013
Depending on your particular environment, database reporting can have a heavy impact on the database performance, can execute queries which run for dozens of minutes or both. This is usually the case with reports which require complex queries having multiple calls to SQL Server’s aggregate functions to be executed against very large data sets. The effect on the database performance is particularly severe in scenarios where the production database stores data which is changed often – as data modification operations require exclusive locks, in order to preserve the integrity of the database, SQL Server will go ahead with the data modification operation until the query initiated by the report is still running. This increases the chances of a deadlock occurring; especially in cases where another set of data modification instructions, dependent on the ones which are waiting for the reporting to finish has already been applied. Therefore, reporting can cause performance degradation in a production environment. So, how can reporting be optimized to prevent such a heavy load on production databases?April 4, 2013
Nobody likes to read a wall of text, even when it’s just plain text. When it comes to reading code, the problem is even bigger. Code can have different formatting styles, which could make a job either easier or more difficult. It can make code difficult to decipher and understand. A clean and neat SQL is read faster than an inconsistently written SQL; SQL reviewing and troubleshooting is more efficient; joint development efforts are more effective; handing off projects from one team to another is smoother.April 4, 2013
Determining just the right primary key for tables is one of the most important parts of a robust, high-quality database design. The key candidates and the keys themselves need to be picked with caution, as suboptimal choices can snowball out of control and leave the bloated, slow databases which require heavy maintenance and require massive amounts of work to meet changes in the business requirements. Therefore, due to the importance of the primary keys for the future behavior of the database, their impact on the database performance needs to be weighted as well. So, from a performance standpoint, should replacing complex natural keys with a surrogate key be considered?April 4, 2013
In most cases, splitting a table into two tables containing different columns is required in order to address database design changes, business requirements or even adding domain restrictions retroactively (for example, isolating currencies in a separate table and enforcing referential integrity via a foreign key to ensure that only valid currencies can be stored in the database). However, splitting tables may actually offer additional performance benefits to the database.April 4, 2013
Database performance is a challenge for every developer and DBA. Even when some improvements are made, there is always a question – is there anything else that can be done.
There might be. Some of the performance improvement techniques are not code related – a proper indexing strategy, adding memory, using different disks for data files, log files, and database backups, using faster disks, optimizing tempdb performance.April 4, 2013