In a quiet office in Tenkasi, a rural town in Tamil Nadu, a small team of engineers at Zoho Labs has been doing something incredible. This team, comprising mostly locally trained talent from Zoho Schools of Learning, has created what could be India’s first GPUpowered database engine. The power of work Zoho Analyticsenabling businesses to process massive datasets up to five times faster than traditional methods.
What distinguishes this achievement is not only the advanced technology, but also the approach that Zoho took to create it. Operating away from conventional technology hubs, the team started with limited knowledge of GPUs, learning everything step by step – new technologies, database optimizationand GPU programming – before building solutions that compete with global standards.
“This is proof that with the right mindset and resources, transformative technology can be built anywhere and benefit businesses everywhere,” said Jagan, head of the GPU team at Zoho Labs.
The meeting is necessary for speed
Zoho’s GPU database engine was developed to address one of the main demands of modern analytics: Speed. Businesses rely on analytics platforms to quickly make sense of data. Traditional methods for database optimization, such as upgrading the CPU or using faster hardware, often limit their performance. GPUs, known for their ability to handle parallel computing, provide an exciting solution to speed up database operations.
The team’s GPU solution powers Zoho Analytics, a business analytics platform, by optimizing database operations like joins, group-bys, and sorting, so users can efficiently extract actionable insights from data. However, building this capability requires overcoming major technical hurdles.
One of them involves managing data transfer between CPU and GPU. Since the GPU relies on dedicated memory, data must be transferred to the GPU memory via the PCI Express bus before processing can begin. This transfer process can become a bottleneck, increasing the execution time. The team developed techniques to minimize this overhead, ensuring faster and more efficient query execution.
The journey also involved designing solutions to specific challenges, such as creating optimized hash tables and sorting algorithms. These innovations allow the system to handle different types of data and increase processing speed for heavy workloads, which are common in analytics.
And while GPUs excel at numerical data processing, they face challenges with dynamic data types like text. The team overcomes this by developing special kernels and algorithms to manage these operations efficiently, making the system suitable for a wide range of data analysis tasks.
A homegrown approach to innovation
Zoho, which has annual revenues of more than $1 billion, builds cloud software for businesses, and has an enterprise IT management division called ManageEngine. Founder Sridhar Vembu established the Tenkasi center in 2011. Today, most of the company’s work takes place here. Most of Zoho’s tools and technology used to build products are developed in-house, and the company’s reliance on its own data centers further reduces costs. This allows companies to offer advanced capabilities like GPU-accelerated analytics at no additional cost to customers.
Zoho’s Schools of Learning program also played an important role in its success. Graduates of this program, who often come without formal degrees, are trained in programming and problem solving. Jagan is a product of the program and it prepares him and his team to tackle complex projects.
“The confidence to approach something new, like GPU programming, comes from a strong foundation in the Zoho School,” Jagan said.
The GPU-powered analytics platform is now an integral part of Zoho Analytics. Customers don’t need to worry about setting up infrastructure or managing servers – Zoho Analytics handles it all.
The adoption of GPUs for database optimization is in line with broader trends in the technology industry. The growing popularity of AI has led to advances in GPU hardware, including expanded memory capacity and faster interconnects. These developments benefit not only AI applications but also analytics platforms like Zoho, which rely on GPUs to gain performance.
AI, Jagan says, has brought greater awareness of GPU capabilities. And advances in hardware are helping them improve their solutions.
