Skip to main content

Improving the upload experience inside an Azure web application


In most of the cases, the latency problems are from the backend to our customers. They want to be able to access our data as fast as possible. Sometimes things are different, and the real challenge is to find ways how you can provide a reliable and fast way for customers to upload their content to the backend system.

Imagine that you have a system where agents around the world upload insurance policies every few minutes. The system is hosted in only one Azure Region – UK West. From the quality attributes of the system, there is no need to deploy the solution in multiple locations and building a native application or plugin for this purpose it is not acceptable (there is no business justification). Even so, agents are all around the world and the latency and internet speed, it’s a fact that needs to be handled.
At a high level, the system uses Azure Storage and Azure CosmosDB to store information that it is uploaded by agents. Agents are using a web portal and some REST API that runs inside Azure Web Apps. A horde of Azure Functions is crunking the uploaded content and push the content over Azure Service Bus to other backend systems that are out of our scope. Every day more than 5000 agents are online and on average, there are 500000 files uploaded every day.

Client requirement is simple:

Reduce as much as possible the latency and improve the upload experience of the agents. They complain that uploaded speed and application reliability is a big concern.


Let’s keep things as simple as possible and accept that deploying the solution in multiple locations it is not acceptable. Azure CDN can be used to cache content when you want to be closer to the customers, but we need the opposite thing.

The set of actions that we can take are:
(1) Improve the reliability of the system that handles the upload
(2) Improve the client experience on the web application
(3) Identify how you can bring the land zone closer to the agents

Upload reliability
The system already uses HTTPS protocol that offers pause and resume of an upload. To improve the performance and reduce the stress on the web application, we could rely on Azure Blob Storage. The agents could upload content directly to an Azure Storage – an upload link together with a SAS could be generated automatically.
From my experience, I notified that the scalability and reliability of uploading content directly to Azure Blob Storage is much higher vs implementing your own API.

Client experience

With the new capabilities that we have nowadays in the browser, we can upload content in the backend with the full capability of resume and do multi-part without having to write to much code. The agent can carry on with other tasks while the files are uploaded. JavaScript libraries enable us to offer a seamless experience in any modern browser. Filepond and Uppy are two great libraries that can be easily integrated inside the system.
The reliability of the connectivity is one of the biggest concerns, especially for agents that are using the system while they commute (e.g. unreliable connectivity when a train is inside a tunnel). In these situations, the client web application shall use the local storage to cache content and persist user information until the connectivity is restored. Combined with a Filepond, the client experience should be the same even if they lose connectivity for a few seconds or minutes.

Bringing file land zone closer to agents
There are multiple approaches to this problem. From the start, Azure CDN is not a viable solution; it is used to bring static content closer to customers and not the other way around. At this moment in time, there are 58 different Azure Regions around the globe. A simple solution would be to create multiple landing zones in different Azure Regions by measuring the latency between agents and the UK region.
After analyzing the latency and the agents' distribution, the decision was the following: Create Azure Blob Storages that are used as landing zones in the following Azure Regions:
  • UAE Central
  • Japan West
  • East Asia
  • North Central US
  • South Africa West
The backend system identifies the user location automatically and allocates the closest landing zone, that it is used by the client web application to upload the content. The agent is not aware related to the location where content is uploaded. Besides this solution, there are also other commercial solutions on the market, based on the same principle.
After doing a PoC we identify that the upload experience is not improved drastically by the solution. Why? Because the real issues were not from the latency as the client expected. It is from the way how the client web application is implemented.

Conclusion
After analyzing the above options, measuring the impact and the outcome, the results were the following.
  • By having multiple landing zones in different geographical locations, there is no real improvement to upload experience. The complexity of the system increases drastically without a real benefit.
  • The highest impact is by improving the client web application, adding offline and stateful capabilities. The latency and unreliable internet connectivity can be easilty overcome using a JavaScript library like Filepond that uploads content directly to Azure Blob Storage using a SAS. These, combined with a simple local database that cache agent data until the connectivity is reliable again has the most significant impact on the system. The selected JavaScript library for handling offline mode is Hoodie because of the strong integration with local and data storage.

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