Skip to main content

(Part 2) Testing the limits of Windows Azure Service Bus


In one of my last post we talked about the performance of Windows Azure Service Bus and what is the maximum number of messages that we could process in a given period of time.
In this post we will see what is the maximum numbers of consumers that we can have for subscription on Service Bus. Before staring, you should know that this refers directly to our own requirements and with our custom actions that we need to do for each message. Because of this, for each kind of application, the result can vary a lot.
Scope:
To see what is the “magic” number of Windows Azure Worker Roles instances that we can have on cloud that can process messages from only one topic.
Environment:
Each message that was added to Service Bus was pretty small. We had around 100 characters in UTF7 and 3 properties added to each BrokeredMessage.
We run the tests with one; two and three subscribers with different filter rules and the result were similar.
Each message that is received from Service Bus required a custom action to be executed. This action is pretty complicated and consumes CPU power. Also the logic requires to access remote services (that are stored in the same data-center).
Action:
First step was to use only one instance that consumes messages. We tried to determine what kind of instance is more suitable for us – from CPU perspective and number of messages that are processed. We tried the size of the instance type that offers us the best price per instance also.
Remarks: The processer level of each instance was at around 80-85%. We use an automatic scaling mechanism over threads that take into account the CPU power, number of processors and so on. During the tests, there were client that produces messages in the same time – this was required because we had to see how Service Bus behave in a real situation, not in a situation when topic already contains all the messages
Second step was to identify the size of instance that was most suitable for us from all perspective. The cost was one of the most important one. After this we scale from 1 to n instance to identity where what is the maximum number of instances that we can have.
Results:
After first step we identify that 100.000 messages can be processed by roles in the following time:

  • Small – ~19 minutes
  • Medium – ~18 minutes
  • Large – ~18 minutes
  • Extra large – ~17 minutes

The best performance related to costs was offered by the Medium size instances.
Next step was to see what is the processing time of 100.000 messages using different number of instances. All the instances that were used at this step is Medium. We end up with the following times:

  • 1 instance – 18 minutes
  • 2 instances – 11 minutes 
  • 3 instances – 7 minutes
  • 4 instances - 4 minutes
  • 5 instances – 3 minutes
  • 6 instances – 2 minutes
  • 7 instances – 2 minutes
  • 8 instances - 2 minutes

In our case we observed that the number of instances that would make us happy and the cost would be pretty okay also is around 4 instances.
Costs:
The costs for processing 1.000.000 messages are pretty good:  4 x 0.16$ + 1$ = 1.64$.
Conclusion:
In this post we saw what is the size of the worker role that is most suitable for us. Using a Medium size worker role will offer us the best performance, reported to price. In the next post we will take a look over the duration of processing different number of messages using Medium size instances and differet number of instances.

Comments

Popular posts from this blog

Windows Docker Containers can make WIN32 API calls, use COM and ASP.NET WebForms

After the last post , I received two interesting questions related to Docker and Windows. People were interested if we do Win32 API calls from a Docker container and if there is support for COM. WIN32 Support To test calls to WIN32 API, let’s try to populate SYSTEM_INFO class. [StructLayout(LayoutKind.Sequential)] public struct SYSTEM_INFO { public uint dwOemId; public uint dwPageSize; public uint lpMinimumApplicationAddress; public uint lpMaximumApplicationAddress; public uint dwActiveProcessorMask; public uint dwNumberOfProcessors; public uint dwProcessorType; public uint dwAllocationGranularity; public uint dwProcessorLevel; public uint dwProcessorRevision; } ... [DllImport("kernel32")] static extern void GetSystemInfo(ref SYSTEM_INFO pSI); ... SYSTEM_INFO pSI = new SYSTEM_INFO(

Azure AD and AWS Cognito side-by-side

In the last few weeks, I was involved in multiple opportunities on Microsoft Azure and Amazon, where we had to analyse AWS Cognito, Azure AD and other solutions that are available on the market. I decided to consolidate in one post all features and differences that I identified for both of them that we should need to take into account. Take into account that Azure AD is an identity and access management services well integrated with Microsoft stack. In comparison, AWS Cognito is just a user sign-up, sign-in and access control and nothing more. The focus is not on the main features, is more on small things that can make a difference when you want to decide where we want to store and manage our users.  This information might be useful in the future when we need to decide where we want to keep and manage our users.  Feature Azure AD (B2C, B2C) AWS Cognito Access token lifetime Default 1h – the value is configurable 1h – cannot be modified

What to do when you hit the throughput limits of Azure Storage (Blobs)

In this post we will talk about how we can detect when we hit a throughput limit of Azure Storage and what we can do in that moment. Context If we take a look on Scalability Targets of Azure Storage ( https://azure.microsoft.com/en-us/documentation/articles/storage-scalability-targets/ ) we will observe that the limits are prety high. But, based on our business logic we can end up at this limits. If you create a system that is hitted by a high number of device, you can hit easily the total number of requests rate that can be done on a Storage Account. This limits on Azure is 20.000 IOPS (entities or messages per second) where (and this is very important) the size of the request is 1KB. Normally, if you make a load tests where 20.000 clients will hit different blobs storages from the same Azure Storage Account, this limits can be reached. How we can detect this problem? From client, we can detect that this limits was reached based on the HTTP error code that is returned by HTTP