Netreo is now BMC. Read theBlog

5 Tips for optimizing applications on Azure for performance

By: Netreo
June 9, 2015

Azure is a scalable cloud computing architecture that is designed to deliver enterprise class IT resources for simple to complex applications. It is home to some of the top applications in the world and handles thousands of instructions per day. Though it is built to deliver supreme operational performance to any application, it still requires a robust architectural design and timely monitoring to ensure an uninterrupted application delivery.

Application and server administrator feels that application performance can be optimized at run time but it pays far more to design application with performance and scalability in perspective. Operational performance is easier to tune than bad application architecture and design.

This blog post will highlight five key Azure architectural design tactics that can help you optimize and increase the performance of your applications on Azure.  These tactics are advised in lieu of the fact that Azure infrastructure is designed to work on shared environment where multiple instances of an application are used to deliver a failover mechanism.

Tip # 1 – All in One Data Center

optimizing applications

It is no surprise that how quick and robust networks can be, if your application’s resources are hosted in different data centers- it is prone to performance lag. Data Center location might not be directly an application design paradigm but wouldn’t it be easier to code, test and manage an application hosted at one central data center?

A perfect Azure Application Architecture will bring all nodes and applications as close as possible- ideally in the same data center. Your applications will receive far greater performance (overall) if you deploy the SQL Database, Servers and all other resources on Azure.

Tip # 2 – Design for minimal Network Latency

Network Latency / optimizing applications

Cloud services platform such as Azure are always prone to network latencies.  End users requests have to pass through a myriad of routers, switches, intermediary networks, ISP before they reach the application server. Such complex network architecture surfaces network latencies, interruptions and common disconnections issues.  

Azure application developers must constantly load and stress test their applications and looks for ways to reduce network latency. The core design and code of the application should be as lean and clean as possible. User request (inbound and outbound) must carry as minimum payload as possible to eliminate network congestion. Having all your apps in one data center will minimize the different networks it has to travel from.


Tip# 3 – Fault Tolerant Design


The application designed must be able to handle dropped connections and quick to react to exceptions. Most of these errors are temporary in nature and caused by conditions such as network connectivity issues or service unavailability. Application developers must utilize ways to enforce fault logic and handling mechanism.

One of the best ways to do this is by using Transient Fault Handling Application Block that adds robust fault handling logic with your Azure stack. By implementing incremental, fixed internal and exponential back off retry strategies, application developers can add fault handling logic to SQL Azure, Azure Service Bus, Azure Storage and Azure Caching Service.

Tip # 4 – Monitor Engine Throttling

Engine Throttling

Azure being a cloud service applies a certain limit on the resources that are consumed. However, if your application exceeds those limits, you are likely to receive a throttling error or experience downtime. Azure SQL Database has two types of throttles – soft throttles and hard throttles. Soft throttles are peak resource usage for a limited period of time. Though they should be monitored too but they are less likely to deliver a service failure error. However, hard throttles are red sign for sure and can cause service failure.

It is important for Azure Administrators to monitor both incidents and report that to application designers / developers. Azure Automation and Monitoring Software can easily track and report on different engine throttling stages. Having such an application since the start will help in optimizing performance from the very beginning.

Tip # 5 – Plan and Design Capacity


Last but not the least, application developers and server administrators must have the eye to perceive workload in times to come. It is an essential performance planning step that trumps all other. If you fail to provide enough storage, memory and processor, your application is most likely to taste the floor. Even if somehow it remains online, the execution and delivery time of tasks and resources will dramatically fall.

And there is no excuse in capacity planning for Azure users.

Azure has eliminated the fatigue old days of waiting for computers and storage to add up in the infrastructure. Azure provides instant access to compute nodes, blob storage, table storage, queues etc. You can scale up and down as needed. Use the Auto Scaling Block for applications that are expected to see a sudden usage spikes. This Block allows you to set rules for scaling per role instances. Application developers and administrators can either enforce constraint rules (rules such as setting a min/max threshold for instances per day) or reactive rules (that occur when a specific condition is met such as memory usage %).

Although there are hundreds of performance techniques and tactics for Azure applications, but this article has broadly described the most essential application performance strategies. Netreo Azure Automation and Monitoring Software simplify the process of monitoring, maintaining and optimizing your azure application and in delivering an uninterrupted user experience.


See how Netreo can ensure top performance for your application.

Image Source in order: cooldesign, jesadaphorn, David Castillo, Giovanni Sades at freedigitalphotos.net

Ready to get started?

Get in touch or schedule a demo

Get Started Learn More