Tagged articles - SQL Server

Caching data using Microsoft SQL Server may not be the most popular way of caching data because of the performance comparing to caching data to Redis, but it is maybe the most convenient way to keep your state of the application outside the process itself. Beside, if you are already using SQL Server database, you can just move the caching infrastructure to a separate schema inside the same database and keep using it until requirements are met for cache storage of better performance (like Redis for example). This makes infrastructure not so complex and less dependent, again until the requirements for something faster are met. ...read more

Since the time of ADO in .NET one of the limitations was that one connection can execute single command at the time. Same limitation Entity Framework inherited and you can execute single command at the time per instance of your DbContext. Typical dependency injection lifetime of DbContext is scoped, which means you can execute one command at the time as the DbContext is instantiated per HTTP request in ASP.NET Core application. ...read more

Not so while ago I wrote an article on Seeding data with Entity Framework Core using migrations. This approach relies on EF Core migrations to ensure seeding of specific data is done only once using __EFMigrationsHistory table to track structural migrations as well as data seeding migrations. In large number of cases this approach works just fine and you do not need to do any adjustments or add any additional tables, just use migrations mechanism out of the box...read more

Object relational mappers or ORMs are a great tool which can significantly reduce application development time, but if they are not use wisely they can cause bottlenecks and performance issues. Same goes for dependency injection containers. The lifetime of the components, especially the ones that can cause recurse locks like SQL Server connections must be as short as possible and they should be released as soon as work is done. ...read more

Some time ago I wrote an article which explains how to Increase service resilience using Polly and retry pattern in ASP.NET Core. This is a great way how to easily implement retrials when using .NET Core dependency injection, but in case of using Autofac with .NET Framework 4.x you do not have many out of the box solutions. However, Polly as a library is not specifically built for .NET Core and using it with other dependecy injection packages such as Autofac is fairly easy...read more

Over the time various techniques to manipulate database data have evolved in .NET. Entity Framework is certainly the most advanced one, but for small ad hock applications where you have few inserts to deal with, ADO is still the fastest way to write data to the database...read more

Unlike Entity Framework 6, EF Core and whole .NET Core relies on dependency injection. For example, configuration is expected to be injected from the calling assembly instead of being available at any time. This is one of the reasons Migrations in Code First approach do not work out of the box for EF Core...read more

JSON is the most commonly used data fromat nowdays. It is lighter than XML, more readable and it's sindax complies with most OOP C-Like lnguages such as Java, JavaScript and C#. There is not really many downsides of using JSON, so that is one of the rasons everyone is using it in one way or another. ...read more

Sending email from SQL is not a difficult to achieve, but generating rich HTML content might be. SQL Server itself is not built as a text processor, first approach would be to build HTML by simple concatenating the string elements. This approach will work, but over time it will become more and more difficult to maintain especially if you need to add layout elements over time. ...read more

So far (based on my experience) the fastest way to insert big number of records from application to database is to use SqlBulkCopy. The downsize of this is that SqlBulkCopy uses DataTable instace as an input parameter which is not so convenient when dealing with models which are strongly typed and a lot more easy to use than iterating through the DataTable. Since SqlBulkCopy accepts DataTable for insert you can not use models directly to insert your data...read more

In case you need to handle IP addresses in SQL, they are not much useful as string values. To do anything with them rather then just storing and retrieving (checking the ranges for example) you need to split IP address to segments (octets). The following is modified method I run into online. ...read more

Probably the most common error developers make in queries is dealing with date ranges and most of the time it is because of missing the minimum and maximum time of the specific date. Month ...read more

In case your data in tables needs to have unique constraint based on more than one value in the row, the proper solution for that would be a composite key. ...read more

Nowadays, in the years of ORM, barely anyone thinks about stored procedures, but they are still there and accessing and reading data with ADO and SqlDataReader ist still the fastest way. Stored procedures are validated, parsed and compiled in SQL server, so these steps are not required to be performed versus plaint text query (which is what entity framework generates for accessing data). ...read more

Entity framework is an awesome tool for working with database in .net, but sometime you have to get back to old fashion ADO for the performances. The bad thing is that as soon as you get back to ADO, you have to deal with SqlDataReader, DataTable and DataSet classes. This is not so convenient for using in upper levels of applications, especially in MVC environment where using strongly typed objects is an essential thing. ...read more