Datonics is one of the world’s leading firms for aggregating and distributing search and targeted digital advertising data. The company was founded in 2011 as a subsidiary of Almondnet, a pioneer company of targeted Internet advertising.
Companies worldwide use its flagship data analytics platform to create and deliver programmatic ad campaigns. It also helps companies process their own data to meet their business goals.
From its earliest years, Datonics had to grow quickly to match customers’ demand for data, so much so it was creating challenges of scale, which led to Datonics CTO Adi Panhasi meeting Vadim Solovey, CEO of DoiT and what would lead to one DoiT’s most enduring business relationships.
Datonics had early success attracting and growing its customer base but faced limitations in processing large datasets quickly and cheaply. It began initially with a co-location hosting service but soon reached a point where queries would take hours to process or fail outright.
Data brokers like Datonics often face the challenge of exponential cost increases just to maintain the accuracy of their large datasets. Left unchecked, this creates analysis problems caused by duplication, difficulty integrating data from multiple sources, and inaccurate predictions, usually referred to as ‘junk inferences.’
This situation was not easily scalable, especially as it was damaging Datonics’ ability to run calculations efficiently for its customers. Dataonics needed to find ways to optimize its infrastructure.
The relationship between the companies began when Vadim met Adi. “I think more than 10 years,” recalls Adi. “I’d have to look through my old emails, and I’m looking at 2013.” In trying to solve its challenges by moving to Google Cloud, Datonics became one of DoiT’s earliest customers.
“In 10 years, I don’t remember a single case where we were disappointed with the service or professionalism of the staff at DoiT. I have only good words to say,” says Adi. “From early on, we both learned each other’s capabilities. Vadim started with carefully understanding where we wanted to go and not just doing the IT work but consulting with us on where we should go.”
Datonics now relies heavily on Google Cloud infrastructure, particularly BigQuery, Google’s serverless data warehouse platform. More recently, the company discovered that even years after moving to the cloud, Datonics had ongoing challenges with the large volumes of data it was handling. The company was looking at much more expensive resources and slots on BigQuery using the on-demand pricing plan, which can easily run into $20,000+ per month due to the size of the datasets they were working with.
Additionally, there were security concerns over the disruption this had on operations and service delivery. This meant delays in customer deliverables and an impact on overall customer satisfaction.
Knowing this, the team at Datonics turned to DoiT to figure out how to balance cloud scalability with cost control so that its infrastructure could continue to grow without an exponential increase in cost.
DoiT takes a consultative approach to working with customers like Datonics. This involves finding a comfortable balance for the customer between providing resources for self-learning so internal teams can handle issues themselves and hands-on guidance, with DoiT’s team suggesting suitable solutions and then implementing them for the client.
DoiT’s Cloud Data Architect, Elad Shaabi, undertook a full investigation to find the root cause of Datonics’ scalability challenges as part of DoiT’s Cloud Native Training. He came back with different strategies to tackle the issue, running various proofs of concept to make sure the solutions would be suitable for Datonics’ cloud environment. After exploring a few options for optimizing their systems, Datonics decided to solve their challenge by implementing a HyperLogLog (HLL) algorithm to speed up query processing time and use significantly fewer resources.
HLL is a data structure designed to assist in making estimations with large datasets without using excessive amounts of memory required by more traditional mathematical techniques. This is an algorithm that Google itself implements internally with BigQuery for its cardinality estimation.
Once DoiT had developed a working proof of concept for Datonics, they were presented with the solution before integrating it into small packages for ease of implementation. With this done, results were reviewed before a final consultation with both teams to make sure they had the results they wanted.
Implementing HLL in Datonics systems significantly improved query performance. Where before queries would sometimes take hours to run, they are now completed in only seven seconds, a 99% improvement.
It also saves Datonics on resource costs, allowing it to run a query once and apply different data dimensions rather than rerun a full query for each dimension. This has cut slot usage from 2,000 to 132, a 93% reduction, while data scanned per query has fallen by 99%.
“DoiT has given us a good balance of giving us directions to the sources, doing our own learning and reading, and providing an opinion on which solution would support us best,” says Adi.
DoiT has become a vital partner for Datonics, operating almost as an extension of its team. Datonics’ team spends less time on troubleshooting, and the optimizations have avoided the need for costly infrastructure upgrades, saving Datonics thousands each month.
Datonics’ long-term goals include continued growth and upcoming expansion into new markets, backed by its decade-long partnership with DoiT as an extra team member who deeply understands its business.
Datonics has found DoiT’s team to reliably integrate with its cloud challenges, with each person quickly becoming familiar with its needs and effectively answering its queries every time.
The cost savings and performance improvements achieved through implementing HLL with BigQuery have been significant. This allows Datonics to manage data volumes previously beyond its budgetary and infrastructure capacity.
Paradox Interactive partnered with DoiT to gain full visibility into cloud costs, improving resource optimization and efficiency. This collaboration, built on trust and transparency, has empowered Paradox’s cloud operations team to make data-driven decisions.
As cybersecurity threats in the automotive industry grow, PlaxidityX is building a next-generation multi-cloud security platform to protect vehicles inside and out.
Discover how FREENOW collaborated with DoiT to optimize cloud costs, and enhance operational efficiency.
From cost optimization to cloud migration, machine learning and CloudOps,
we’re here to make the public cloud easy.
From cost optimization to cloud migration, machine learning and CloudOps, we’re here to make the public cloud easy.
Ready to get started?
You will receive a calendar invite to the email address provided below for a 15-minute call with one of our team members to discuss your needs.
