/Taxilla_logo%402x.1ce909221a95c74271b837fb9f7ea99f31dbf3e8.png "Taxilla")
Taxilla Runs on AWS Graviton Processors to Halve Data Processing Costs
2022
After India introduced the Goods and Services Tax (GST) in 2017, the compliance burden—much of which requires online, electronic filing—significantly increased for businesses operating in the country. Furthermore, digitization has been highlighted as a GST trend going forward, with
technology-based tax administration gaining greater importance. Now more than ever, companies need a modern digital solution to efficiently meet evolving regulatory requirements.
Taxilla is a compliance and digital transformation platform launched in 2017 that helps businesses in India prepare and process data. Its services also include document submission to government authorities and B2B compliance processing for businesses with various internal reporting requirements. Over 300 mid-sized and enterprise customers in India currently use Taxilla applications such as enComply, a GST compliance tool and the first product that Taxilla launched in its application suite.
Launching New Products Causes Traffic—and Costs—to Rise
Taxilla was born on the Amazon Web Services (AWS) Cloud. The company chose AWS for its maturity in the Indian market and reliable infrastructure. The Taxilla platform also relies on open-source technologies such as ElasticSearch, Java, Apache Kafka, Apache Spark, Apache Flink, and Cassandra, which run on
Amazon Elastic Compute Cloud (Amazon EC2) instances.
Toward the end of 2020, the company started expanding into the Middle East and launched enInvoice, an application to help businesses automate e-invoicing. Traffic volumes on the Taxilla platform rose sharply with the addition of enInvoice. Moreover, customers started using the platform in a fundamentally different way.
Santanu Saraswati, CTO at Taxilla, explains, “With enComply, we were supporting customers’ periodic compliance needs. But with enInvoice, the use case changed to continuous compliance, so customers were using our services throughout the day, 24x7.” Cloud infrastructure costs spiked with the constant increase in user traffic, which led the firm to begin a cost optimization journey. Taxilla consulted AWS on the best way to support both the enComply use case for bulk/batch data processing and the enInvoice use case for continuous, real-time processing.
With Graviton, we gained a 50 percent improvement in price and performance versus our previous x86 processors.”
Santanu Saraswati
CTO, Taxilla
Catering to Highly Performance-Sensitive Workloads
As a first step toward reducing costs, Taxilla adopted
Amazon EC2 Spot Instances in its non-production environments to optimize Amazon EC2 costs. It then began testing instances with different types of processors, namely x86-AMD and Arm-based
AWS Graviton processors, to support production workloads that were previously powered by instances using x86-Intel processors.
Saraswati says, “Adoption of Amazon EC2 Spot Instances and Graviton processors were completely complementary and didn’t have any dependencies. We were also keen to explore the synergies of using Graviton with Spot.”
In addition to cost savings, maintaining performance was a key requirement, particularly for enInvoice customers. “E-invoicing is highly performance-sensitive; if the invoicing process stops suddenly or slows down, it could negatively impact our customers’ businesses,” shares Saraswati. “Whatever action we took to manage expenditures could not affect response times or overall performance.”
Gaining Superior Price-Performance
Taxilla first tested multiple workloads on instances powered by x86-AMD processors, which are compatible with x86-Intel processors but at a lower price. During testing, however, engineers noticed that response times were slightly slower for some workloads with x86-AMD processors. This was not the case for Graviton. Workloads tested on Graviton matched the performance targets that Taxilla experienced using Intel processors, which was a key differentiator. The price reduction from Intel to AMD or Graviton was similar, engineers found, but the company ultimately chose Graviton for its superior price-performance
“With Graviton, we gained a 50 percent improvement in price-performance versus our previous x86 processors,” Saraswati says. “Effectively, our AWS bill returned to the same level it was before enInvoice was released.” Taxilla also experienced a 10 percent performance improvement in bulk processing and 28 percent faster transactional processing with Graviton.
The success that the company experienced with Graviton persuaded Taxilla to migrate its full production stack, in addition to its development and UAT environment, from Intel to Graviton processors. “If the instance types we need are available, we are giving priority to Graviton for existing and new workloads,” attests Saraswati.
Receiving Support for Custom Software Builds
When Taxilla adopted Graviton processors, it had to find an Arm build for open-source software used to run its platform such as Elasticsearch, Apache Kafka, Prometheus, and Cassandra. Many of the software providers that Taxilla worked with provided Arm builds in addition to standard x86 builds, but in cases where they didn’t, Taxilla received support from AWS solutions architects.
“Arm seems to be the primary processor that will drive cloud-based applications in the future because of the price and performance benefits. After we resolved the custom build requirements of some software (e.g. Prometheus - which did not have an arm64 build at that time), all our products, base infrastructure, and dependent software are running well on Graviton. For companies running a lot of open-source software, the migration to Graviton is easy,” Saraswati relates. He recommends that other companies embarking on the Arm transition journey with AWS avoid processor-specific software configurations.
Building a Cost-Efficient, On-Demand Big Data Pipeline
Recently, Taxilla added another big data processing application to its platform, enReconcile. Customers use enReconcile to compare data across systems within an organization and discover deviations that indicate data entry or calculation errors. By using enReconcile, customers can save time and effort during auditing and avoid statutory penalties for accounting mistakes.
Taxilla uses
Amazon Elastic MapReduce (Amazon EMR) for cost-efficient data processing in the enReconcile application, in conjunction with Apache Flink clusters running on Graviton processors. “Amazon EMR is a critical component in our big data pipeline because it supports real-time and batch processing on demand and at any scale,” Saraswati says.
Converting to Graviton for New Application Stack
Currently, Taxilla is exploring the use of Graviton for the entire enReconcile application stack, which comprises CPU-intensive use cases. “So far, all the use cases we’ve tested have been well catered for on Graviton,” shares Saraswati.
In addition to new product launches, the business is also planning an expansion to the US in the near future. Saraswati concludes, “All the software and services we’re using on AWS are available in the regions we’re targeting, which ensures that we can maintain the same level of service and product offerings in every market.”
Learn More
To learn more, visit aws.amazon.com/ec2.
About Taxilla
Taxilla is a compliance and digital transformation platform offering applications that help businesses stay ahead of evolving regulatory requirements. It has over 300 mid-sized and enterprise customers in India and the Middle East and is currently expanding to the US.
Benefits
- Cuts costs by 50% while maintaining performance
- Accelerates bulk processing times by 10%
- Accelerates transactional processing times by 28%
- Ensures zero downtime or delays in invoice processing
- Offers support for custom Arm software builds
- Enables cost-efficient, real-time big data processing
- Runs open-source software seamlessly
- Supports expansion into new geographic regions
AWS Services Used
Amazon Elastic Compute Cloud
Amazon Elastic Compute Cloud (Amazon EC2) offers the broadest and deepest compute platform, with over 500 instances and choice of the latest processor, storage, networking, operating system, and purchase model to help you best match the needs of your workload.
Amazon EC2 Spot Instances
Amazon EC2 Spot Instances let you take advantage of unused EC2 capacity in the AWS cloud. Spot Instances are available at up to a 90% discount compared to On-Demand prices.
AWS Graviton processors
AWS Graviton processors are designed by AWS to deliver the best price performance for your cloud workloads running in Amazon EC2.
Amazon Elastic MapReduce
Amazon EMR is a cloud big data platform for running large-scale distributed data processing jobs, interactive SQL queries, and machine learning (ML) applications using open-source analytics frameworks such as Apache Spark, Apache Hive, and Presto.
Get Started
Companies of all sizes across all industries are transforming their businesses every day using AWS. Contact our experts and start your own AWS Cloud journey today.