AGL is one of Australia’s largest and oldest energy companies in Australia.
For years, the company has leveraged technology to improve productivity and enhance the overall service offering to customers. Recently, AGL Energy was keen on how the company could further improve its current processes. The answer is utilizing Azure and its services that caters to modern business operations. Here is how the company managed to save over $2 million by doing so.
Companies, particularly large-scale ones, require many tech products and services to operate effectively. Some of these require constant resource usage. In many cases, these get charged at pay-as-you-go rates. Azure Reservations focuses on reducing these types of costs. With Azure Reservations, users can commit to one-year or three-year plans for multiple products. Essentially, this would reduce your resource costs up to 72% on pay-as-you-go prices.
Similarly, adopting Azure offers additional pricing benefits. For example, Azure Hybrid Benefit allows users to take advantage of existing on-premises Windows Servers and/or SQL servers while users migrate to Azure. If you run Windows Server VMs in Azure, users will only get billed at base compute rate. In other words, eligible customers will pay reduced rates for Azure Virtual Machines. This reduced rate also applies to Azure SQL Database and SQL Server on Azure Virtual Machines.
For Electrons Unlimited, it was important to identify the longevity of the reservations and distribute them among different groups. This resulted in $240,000 in savings for the company. Additionally, the Azure Hybrid Benefit pricing offer cut down costs further by $115,000.
AGL also engages in data science as well. This means the company heavily uses data warehouses. But data warehouses at a massive level can sometimes be inefficient, riddled with scaling issues. For example, take a situation where a data warehouse operates 24/7. Actual effective usage may be only during the weekdays. There is a clear resource wastage during weekends. Thereby, these data warehouses would ideally need to function according to the specific resource requirements.
In Electrons Unlimited’s case, it was a matter of recognizing data warehouse usage patterns. Next was an implementation of a “pause un-pause pattern” so that data warehouses function only when required. The process saved $130,000 for the company.
Scaling issues also came in the form of Azure Cosmos DB, a globally distributed, multi-model database service from Microsoft. Optimizing the process required implementing a dynamic scaler. Essentially, this would help Request Units (RUs) scale up to handle bigger loads and level down when not needed. This solution recorded a $94,000 in savings for Electrons Unlimited.
Another component that goes into Electrons Unlimited’s processes include Azure Event Hub. Simply put, this is a big data streaming platform and event ingestion service. As Microsoft states, “Data sent to an event hub can be transformed and stored by using any real-time analytics provider or batching/storage adapters.”
One of the features of Event Hub is the ability to capture your data in near-real time. This feature scales up automatically via Event Hubs throughput units. Unfortunately, it does not scale down in instances where demand levels are low. Furthermore, this capture feature can be an expensive endeavor for non-production workflows.
As such, the Kodez team implemented a scale-down feature to tackle the scaling issue with Event Hub. Additionally, the capture feature was also turned off for most non-production Event Hubs. The process saved up $88,000 for Electron Unlimited.
Speaking of non-production workflows, Azure Dev/Test Subscriptions offers a convenient mechanism to run their development and testing workloads on Azure. Some of the advantages of going with the subscription includes,
Opting for Azure Dev/Test subscriptions helped AGL Energy save a total of $135,000.
Virtual Machines are Electrons Unlimited’s biggest contributor for excessive costs. The use-case for AGL is somewhat extensive due to the company’s size and the complexities involved. Thereby, it was important to identify underutilized VMs to align usage with actual requirements. Owing to the complexity involved, the Kodez team followed a semi-automated process to identify these VMs.
There was also the case of unwanted VMs in idle. Usually, deleted VMs do not atomically clean up the disks as well. There were such several disks that were in idle over time. The solution was to automate the process of contacting the owners of these disks and mark them as delete-able.
Deleting unwanted VM disks saved $420,000 for Electron Unlimited. Meanwhile optimization of the VMs in use resulted in $480,000 in savings.
One of the main areas that needed addressing with regards to storage was switching non-production storage accounts from Geo Redundant Storage (GRS) to Locally Redundant Storage (GRS). In case you are unfamiliar with the concept, LRS copies data three times within a single physical location in the primary region. GRS on the other hand, “copies your data synchronously three times within a single physical location in the primary region using LRS. It then copies your data asynchronously to a single physical location in a secondary region that is hundreds of miles away from the primary region.”
Simply put, GRS retains more copies of your data in your company. While important, this is not a necessity for every work environment. Thereby, it can turn out to be a waste of resources and an expensive exercise. Hence, switching from GRS to LRS for non-production storage accounts. The Kodez team also deleted unwanted files and implemented retention policies to keep the clean-up process on-going. This saved AGL a total $170,000.
The Kodez team’s work also included services like App Services, Databases, API Management Services, Batch Services, and IP addresses. These marked a total savings of $150,000.
In the end, AGL saved up $2,022,000 thanks to many optimizations of its backend services. So, in case you’re curious about how Kodez can help optimize your business do check us out. You could also call us on (+61) 416192293 or email us at firstname.lastname@example.org for details.
Azure MachineLearning Service provides four main compute options each with a specific purpose attached to it. In this post we will go through each of those and see where we can occupy them in your ML experiments.
In the era of Industry 4.0 where data and predictive analytics players the major role, developing machine learning pipelines have become an essential in intelligent application development.
Since instant messaging applications being popular among the users, embedding conversational agents or the Chatbots with such services are essential for better user experience.