I use columnstore indexes on SQL 2014. What would be the advantages of upgrading to SQL 2016?

I use columnstore indexes on SQL 2014. What would be the advantages of upgrading to SQL 2016?

SQL Server 2016 really represents a drastic change for the better in regards to general performance. And the proof is that, thanks to the engine changes, this is the first SQL Server version where Microsoft actively provides a widespread increase in performance levels by simply upgrading. But what impact can it have in your daily loads if you use columnstore indexes in SQL server 2014 datawarehouse? There was already an amazing increase in performance levels with SQL Server 2014 and the columnstore indexes. Is it worth upgrading to SQL 2016? (more…)

Virtual SANs and the trend towards SDS

Virtual SANs and the trend towards SDS

Lately, we are noticing a trend whereby traditional SANs are increasingly losing ground. This can be due to various reasons, such as forced migrations to the cloud, limited budget environments or even customers who are getting annoyed with their storage providers’ draconian terms and conditions. A common example is to find very costly maintenance agreements, higher-than-market disk size upgrade costs, as well as difficulties to expand/ improve hardware without having to completely replace it. There is also a strong trend to continue going virtual and trading servers like they were commodities. All the above is therefore contributing to an increase in this trend towards SDS, while the VSANs continue to attract new supporters.


The SQL Server Browser service and multiple instances

The Browser service is a service that can be confusing if you don’t know how it works and what it does. You may have noticed that it sometimes is activated and sometimes not, even if you run the same OS and the same SQL Server version on two identical servers. The answer to this slightly confusing behavior is that the Browser service only gets installed when you create a named instance. If you created a default instance, the service will be disabled.

The functionality that the Browser provides is that it listens on UDP port 1434, and replies with the correct port for the instance requested. For example, if I connect to MyServer\MyInstance, a udp call to port 1434 is done and the Browser service replies with the correct TCP port number for my client to use when it does the “real” connection to the instance. The TCP port in use can be either fixed or dynamic depending on firewall rules on the network and other customer specific rules.

From Microsoft TechNet: Upon startup, SQL Server Browser starts and claims UDP port 1434. SQL Server Browser reads the registry, identifies all instances of SQL Server on the computer, and notes the ports and named pipes that they use. When a server has two or more network cards, SQL Server Browser returns the first enabled port it encounters for SQL Server. SQL Server Browser support ipv6 and ipv4.

Ok, so the browser will find all my instances. So what if you mix the default instance and named instances on the same server? Since you have named instances, the Browser service will be activated, and according to the text above, the Browser identifies all instances and their ports on the computer, right?


The fact is that the Browser ignores your default instance totally, it only works with named instances. As long as you don’t change the default TCP port 1433, you will however not notice this, since your connection just bypasses the Browser anyway. It’s only when you change port for your default instance that this effect becomes visible. The only way to connect then, is to manually supply the correct port number in your connection string.

This has the practical effect that you really should think twice before using a default instance on a server where you have fixed ports and named instances. Let´s say you want to consolidate three SQL Server instances into one server and they need to have fixed ports because of firewall rules. You can either create one default instance and two named instances, or create all three as named instances. If you create all three as named instances, the Browser will deliver their TCP ports without problems, your connection string will simply be MyServer\MyInstance(1,2,3). If you choose the setup with one default instance which is not on port 1433, you will have to use a connection string like MyServer,1444 for your default instance, and then MyServer\MyInstance for the other two. This is not only ugly and non-consistent, but also poses a problem in the future if you want to change the fixed port numbers.

Fixed some SQL Server Partition Management Utility Bugs

The SQL Server Partition Management Utility (http://sqlpartitionmgmt.codeplex.com/) is one of the best tool used to manage the partition-switch operations. It is a command line tool and can be integrated in a SSIS package or used to generate the T-SQL scripts needed in a regular “sliding window” partition management scenario. A blog post that shows how to use this tool is this one.

In my case, I wanted to speed the loading of a big partitioned fact table through a SSIS package (that calls two child packages). So this package calls more instances of the tool in order to load more than one staging table in parallel. Each staging table is related to a fact table partition. After each staging table is loaded, the SSIS package loads the target fact table using the partition-switch operations against the staging table.

All seemed to work fine, but during the test phase, when I tried to increase the degree of parallelism (that is the number of executed instances of the tool), I got a deadlock error.


String Replacement Puzzle

Courtesy of SQL Server Pro. You can find the original article here.

My dad, Gabriel Ben-Gan, passed away recently. He loved numbers, logic and puzzles, and used to solve problems in his own unique way. This article is about a puzzle that incorporates the above ingredients. Dad, this one’s for you, and is in your memory.


Automating software development

One of the things I am fond is reducing the time to perform tasks in the software world. Sometimes that involves lots of code such as a custom testing system I am building for a client. But even lots of the code for the test system can be automated.For instance, consider the application that’s using SQL Server on the backend. The following two select statements spit out labels and textboxes for a WPF window:


SQL Server 2014 – Intense read/write contention (3/3)

In Part I of this series, we reviewed the benefits that In-Memory OLTP may provide for workloads that are bound by insert performance.In part II, we took a look at the benefits it may offer for workloads that are less constrained by inserts, but require ultra-fast access to read-only, or read-mostly data. Today, we will look at the benefits that In-Memory OLTP can offer workloads that suffer from extreme contention between reads and writes.