Skip to main content

Part 2 - Overengineering of a cloud application

In the last post we looked over a cloud solution design to ingest small CSV files uploaded by users. This files were crunched by the system that would generate static reports based on the content. Nothing fancy or complex.
The NFR requirements are light, because the real business value stays in the generated reports:

  • Under 200 users worldwide 
  • Concurrency level is 10% (20 users online simultan) 
  • Less than 15 CSV uploaded in total per day 
  • Basic reporting functionality 
  • Current DB size 150MB (2M reporting entries) 
  • DB forecast for next 3 years is 1GB (20-25M reporting entries) 
  • CSV has up to 1000 entries (maximum 10 columns)
The system that was design for this application was a state of the art system - salable, robust, containing all the current technology trends. But of course was over engineering, to powerful and to expensive.
Now, the biggest concern was how we can reduce the running cost of the system with a minimal impact (development cost). One of the drivers was that we had to come up with a solution that would keep the monthly cost under a specific threshold and also keeping the development cost limited to a specific number of days - rewriting the system was out of the context.

The changes that were done to the system to reduce running cost with a development budget constraints were:
  1. Remove Azure HDInsight - there was no need such a powerful system to analyze 15 CSVs/day (under 15MB)
  2. Remove DocumentDB - reporting data for charts can be stored anywhere. For reports that are static, with a low number of points where updates are happening not very often any kind of storage is perfect. 
  3. Add Power BI - Used as a service it is hanly and useful solution for data visualization. There is full support for CSV files. Once you import them to Power BI you can generate any report on top of them. For 200 users there is no need for 200 Power BI users. The reports can be embedded to a web page. We only have 10 Power BI users in total (5 real users, 3 for pre-production, 2 for testing) that will be reduced to 6 in the near future.
  4. OneDrive - The client already had Office 356 subscription activated, it was a perfect location to store data and fetch them to Power BI.
  5. Remove Azure SQL - There is no need for Azure SQL anymore.
  6. Use Azure AD for user management - The client already user Office 356 an had their location AD federated.
  7. Remove Azure Service Bus - I'm a big fan of messaging system and ESB, but for this context there was no real need. Azure Service Fabric offers internal some 'internal' collections that can replace ESB for this specific case.
  8. Remove Reporting Service - There is no need for this service anymore, Power BI has full support.
  9. Remove CVS Pusher Service - There is no need for such a service. CSVs are uploaded directly to OneDrive and ingested based on their locations
  10. Remove UserMng Service - No need for such a service anymore. IT manage users access based on AD roles and groups.
The running cost were reduced with more than 60% (most of the cost was generated by Azure HDInsight that was a to powerful solution for this kind of 'airplane'. 
We can write a system in multiple ways, using different services and components. The thing that we need to remember that in the end we need to resolve a business problem using a limited number of resources. 

Comments

  1. Nice simplification. I have seem many times, cloud architecture is revolving around running cost and developers tend to use free/low cost services which can be easily done with costly service. eg: someone tried to make documentdb transnational with try catch and background cleanup logic.. which could have easily done with transactional azure sql.

    ReplyDelete

Post a Comment

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