How to monitor your SQL Azure databases sizes

The Web edition of SQL Azure has a hard maximum limit of 5GB, once you hit that limit the server will stop accepting any writes and your application may well go down. It’s important to make sure you have an understanding of what sizes your databases are and how fast they are growing.

At NewOrbit we needed a way to view all our 80+ databases many of which many are dynamically created across multiple subscriptions. Unsurprisingly, it was Powershell to the rescue. There are plenty of weird things about Powershell, but it’s ability to work with XML is fantastic.

So first let’s get our configuration information so we can connect to Azure

	<Data>
	<Subscription>
	<Name>Company1</Name>
	<SubscriptionId>F412A38C-7C34-4763-83E9-001F3DA1E470</SubscriptionId>
	<Thumbprint>4FDS5A6DF1S56D1SA56D1SAD56SA1DS5A6D0S565</Thumbprint>
	<Cert>cert:\CurrentUser\My\</Cert>
	</Subscription>
	<Subscription>
	<Name>Company2</Name>
	<SubscriptionId>FB0BF5DD-F187-4A31-92CD-34DCAE1CFBD7</SubscriptionId>
	<Thumbprint>2DE3442SDSADF5651518161SDA156D1SA56DS0</Thumbprint>
	<Cert>cert:\CurrentUser\My\</Cert>
	</Subscription>
	</Data>

view raw

gistfile1.xml

hosted with ❤ by GitHub

Next, lets loop through our data setting the azure subscriptions. Powershell lets us just use a . notation to walk through nodes. We grab the data for our subscriptions and set our subscription using Set-AzureSubscription. This means that you will need to have the certificate installed on the machine that you are running the script from.

	[xml]$data = Get-Content -path "C:\temp\Subscription.xml"
	$Subs = $data.Data.Subscription
	foreach($Sub in $Subs)
	{
	$thumbprint = $Sub.Thumbprint
	$myCert = Get-Item ($Sub.Cert+$thumbprint)
	$subID = $sub.SubscriptionId
	Set-AzureSubscription -SubscriptionName $Sub.Name -SubscriptionId $subID -Certificate $myCert
	Select-AzureSubscription -SubscriptionName $Sub.Name
	$Servers = Get-AzureSqlDatabaseServer
	foreach($Server in $Servers)
	{
	getServerData $Server $Sub.Name
	}
	}

view raw

Azure Subscription Loop

hosted with ❤ by GitHub

And finally create a function which connects to the Azure Server and loops through the databases (except Master) calculating the percentage of the capacity used

	function GetServerData($Server,$Name)
	{
	$Databases = Get-AzureSqlDatabase -ServerName $Server.ServerName
	foreach($db in $Databases)
	{
	if($db.Name -ne "Master"){
	$PercentUsed = ($db.SizeMB / ($db.MaxSizeGb * 1024)) * 100
	}
	}
	}

view raw

Sql Server loop

hosted with ❤ by GitHub

Once you have access to the databases, there is plenty more you can do including check the database type (Web or Business) and even increasing the size and type  automatically using Set-AzureSqlDatabase

Next we will look at using Powershell dynamic objects to sort and export our data so we can find which databases are closest to the maximum capacity

 

 

Syncing Outlook 2010 and Google Calendar

I use Outlook 2010 at work and Google Calendar at home and on my phone and found after upgrading from Outlook 2007 that the sync tool that Google have only works with Outlook 2003/2007.  I decided to write my own Outlook 2010 VSTO add-in which will sync my entries every hour for the past couple month and three months in the future.

I have uploaded the inital code to github and you can find more details and get the source here http://github.com/MrKevHunter/Outlook2010GoogleCalendarSync

The add in allows you to specify your account details and the number of months of historical and future appointments to sync. When the add in loads for the first time you will be prompted by a settings box



After entering your details and clicking Save your details we be kept in the user.info file with the password encrypted. The add in will then sync your Calendar’s every time you login. I have plans to make it much less garish when I get chance.

You can make the application sync at any time by pressing the sync icon, and change the settings by clicking the settings icon

The code is pretty rough at the moment without enough error handling or unit tests, but that will change when I get time. My testing has found the synchronisation to work okay so far but there is no real logic on matching moved entries (yet), but as we know version 1 sucks anyway

Please don’t hold me responsible if it deletes all your entries, creates crazy new appointments or wipes your hard disk and if you find it useful let me know. Features and problems can be logged on github or in the comments here. I plan to write a bit more about how I wrote it and some of the technologies used (StructureMap, MSpec) when I get chance.

How to query your iTunes library

As much as I hate iTunes its pretty much a necessary evil if you have a iPod or iPhone. One of the biggest problems I have is when a track is moved, there is no way to automatically locate the track, its just marked as missing and there is not much you can do about it other than locate it manually.

I had a browse today and came across a suggestion on Digg for using the WSH and the COM components to find tracks in the library and delete them. I decided to port this over to c# so we could really query the library. Here’s how I did it.

In a Visual Studio project, add a reference to the iTunes type library then create a class and add a method with the following code

public IEnumerable<IITFileOrCDTrack> GetITunesTracks()
{
    foreach (IITFileOrCDTrack track in new iTunesAppClass().LibraryPlaylist.Tracks)
    {
        if (track.Kind == ITTrackKind.ITTrackKindFile)
        {
            yield return track;
        }
    }
}

we now have a query object holding the entire media library. This gives us the ability to find all the tracks in the iTunes library which are missing by writing a simple method such as

public IEnumerable<IITFileOrCDTrack> GetITunesTracksWithoutALocation()
{
    return from track in GetITunesTracks()
    where string.IsNullOrEmpty(track.Location)
    select track;
}

the track object has a delete method which means we can just call delete if we wanted to remove all the missing items from your media library.

This is a quick and easy way to delete lost items, but could also be used for deleting podcasts older than a certain period of time, and probably many other things, hope you find it useful. I can make the source available if anyone needs it.

Finding the best target for refactoring

We are doing a lot of refactoring at the moment modifying systems to do with the PCI requirements of credit card data storage. Part of my job has been changing existing systems to comply with the requirements and we have taken advantage of this to make some tidy up some of the code base. The two ways I have been choosing my targets had been to use NDepend to find its targets using CQL (More about that in another post) and just simply looking at the number of using statements in the file.

Picking  the classes with the most using statements  is a useful way to find the best targets for refactoring.

Getting bitten by enums

One of the issues I have had to fix recently had me scratching my head, and promising myself be more careful when using enums in the future.

For the purposes of this example assume you have an employee class you may decide to store the gender an enum, it will restrict the choices the other developers have when creating or modifying an instance and provides intellisense. When you request an Employee object from your service layer you can then simply check the value of the enums to get the value.

In a nutshell the problem I had was that the information in the database was correct but was not setting the value of the enum. I spent a day of so working through the various layers of the application trying to find what was overwriting the value of the variable until I came to the realisation that the enum had in fact never been set.

What this led to was the system working as expected for a majority of the time and catching us out in a big way. In smaller systems this may not be an issue but when you have lots of layers in your application (too many perhaps?) its worth checking.

The Joys Of Interviews

In last couple of weeks we have been interviewing new contractor candidates. The standard has been pretty good with a few notable exceptions;

1. If you say you find multi-threading easy yet do not know what the lock statement does I tend not to believe you are any good at it.
2. If you have six years experience in c# and describe your competence as excellent, you really should know the difference between abstract and sealed.

People who need to revise for a simple OO test make me shudder.

Hiring again

We’re hiring again at work, getting some contractors in to help us with the deliverables in the next three to six months. From the last time we were interviewing the standard seems to be significantly higher, which I imagine is due to not as many companies taking on contractors. I have heard about people despairing over the standard of .Net developers but I have noticed that many more seem to have an understanding of mocking, testing and the SOLID principles than even a year ago.

So many CV’s came in we changed our hiring process and asked sent the applicants a single method to improve maintainability and email us the results, the standard was pretty high (although the original code was really bad) but I think it will give us a good starting point to discuss with the candidates we get in for interview.

As someone involved in the hiring process we can now afford to be more selective, I have found with our current hiring process we have a reasonably simple multiple choice OO which as perhaps a little too much trivia and a very simple practical which I don’t think that this is enough to really understand how good the candidate is, I would rather that they thought more rationally about design and maintainability than knew their way around SqlCommand and the SqlDataReader.

For any candidate we interview, they should understand unit testing, IOC, separation of concerns, know why large classes and methods are wrong and expect to write code in the interview, not being able to write code under pressure is not a reasonable response. We won’t give them FizzBuzz but might ask them to solve a problem on the whiteboard.

Lastly, if I am interviewing you, one last piece of advice, don’t suggest that I use DataSet’s in my Domain Objects we have enough of that sort of architecture here already!

What’s your interview process and has it changed over the last couple of years?

Can you still be a single platform developer?

When I joined my current company two years ago, I joined specifically as a Winforms developer. At that time the company had seperate Winforms, Web and Service teams within the development department. Within a year this had changed into one cross functional team working on various projects, so I am now exposed to more web programming (ASP.NET 2.0/3.5)  than I expected.

This is definitely a good thing and although I have struggled with some of the challenges of working with a stateless model I am getting to grips with writing the code and getting my ideas out there. I feel a more rounded developer and a more useful team member.

But I was thinking today about if people are still ‘web only’ or ‘winforms’ only developers I think even as little as two years ago that may have been the case. Now I am not sure that you can afford to be. Or if you are, you’re much better off on the web side of the fence. That said, I can never imagine telling a current or prospective employer that I would not be willing to work on a technology even in good times.

I suspect the answer is, and this is from no scientific reasoning and a few Stella’s, that while you still can be a single platform developer, you probably shouldn’t be. The Pragmatic Programmer tells us to learn a new language every year, with the amount of new technology that comes from Microsoft every year perhaps you may just need to learn a new platform.

definitely