If you’re a .Net web developer today, no doubt you’ve enjoyed watching Windows Azure grow up over the past couple of years. The platform has scaled, stabilized (mostly), and added on a slew of great (and sometimes overdue) features. What was once just an endpoint to host a solution, developers today have tremendous flexibility and options in the platform. Organizations are building new solutions and offerings on the platform, and others have, or are in the process of, migrating existing applications out of their own data centers into the Azure cloud.
Whether new application development or migrating legacy, every development shop and IT organization need to monitor their applications in the cloud, the same as they do on premises. Azure Diagnostics has some capabilities, but what I constantly hear from users is that it’s either (a) not enough, or (b) too cumbersome to set up.
Today, Stackify is happy to announce that we fully support Azure deployments, just the same as your on-premises deployments. Let’s take a look below and compare and contrast the options.
Windows Azure Diagnostics
Let’s crack open the Windows Azure documentation on Azure Diagnostics and see just how easy it is to use. The high-level steps are:
Step 1: Import the Diagnostics
Oh, I’ve already deployed my app without the diagnostics module. Guess I can’t do anything until I do this and re-deploy.
Step 2: Configure the Diagnostics (and multiple sub-steps)
Do I want it all? Or just pieces of it? Whoops, forgot to include a specific performance counter, I guess I’ll have to deploy again. Wait a minute… I have to specifically code these performance counters into my role’s OnStart() method, compile and deploy again? And query and consume it myself?
Step 3: (Optional) Permanently store diagnostic data
Lucky for me, Azure storage has gotten pretty cheap. But how often should I move the data into storage? I want to see real-time data, so I guess that’s out now as well.
Step 4: (Optional) View stored diagnostic data
Optional? Of course, I want to see it. Conveniently, Microsoft recommends 3 tools to do this with. Un-conveniently, none of these are web based and they all just give you access to raw data, and very little charting or real-time intelligence. Just….. data. Nevermind that one product seems to have gotten stale since a recent acquisition, and doesn’t even have screenshots!
So, let’s summarize: lots of diagnostics data are available, but think realistically. Think DevOps. What happens when you are in the middle of a major production performance issue and you don’t have the diagnostics you need? You are redeploying an application (and thankfully you have a great branching strategy, so you feel perfectly safe just willy-nilly launching code into prod, don’t you?) to get data, then shipping it to storage, and then digging through that data to find a needle in a haystack. Would you like to be able to troubleshoot a performance issue in the middle of the night, or on a weekend, from your iPad or home computer’s web browser?
Forget it: the best you get is this sparkline in the Azure portal. If it’s real pointy, you probably have an issue; but since there is no alert based on a threshold your customers have likely already let you know. And high CPU, Memory, I/O, or Network doesn’t tell you anything about where the problem is.
The Better Way – Stackify
Stackify supports aplication and server monitoring in real time with our APM tool, Retrace , all through a great web interface. All of the things that Azure Diagnostics provides, Stackify provides for your on-premises deployments, and you don’t need to know ahead of time that you’ll need it. It’s always there, it’s always on. Azure deployments are essentially no different than on-premises. It’s a Windows Server (or Linux) in the cloud. It’s behind a different firewall then your corporate servers. That’s it. Stackify can provide the same powerful tools to your Azure deployments in two simple steps and monitor your azure apps.
Microsoft has shown, consistently, that they love developers, developers, developers. What every developer needs to realize from this is that they’ve given you a canvas, which is exactly what Azure is. It’s great infrastructure that is readily available, easy to manage, and fairly cost effective. However, the tooling is your responsibility. What you get, at best, is bare bones.
App and server diagnostics should be available when you need them. While we, as developers, try to plan for and think of everything ahead of time, there will come times where we need to get data that just isn’t available. And having to go through a lot of cumbersome steps to get that data, and then have to find a friendlier way to consume it…. well, that just doesn’t make a lot of sense to me. I’d rather spend my time writing and developing features and completing bug fixes for my applications than to be writing code to monitor and diagnose.