Scylla Helps Cutting-Edge AD Platforms Scale, Reduce Hardware Costs and Meet Stringent Slas
ScyllaDB, the real-time big data database company, announced that an increasing number of leading advertising technology providers now rely on the Scylla NoSQL database to power their real-time platforms. Ad-tech companies using Scylla include AdGear Technologies, adsquare, AppNexus, GroundTruth, GumGum, Instinctive, MediaMath, Mediarithmics, Numberly (1000mercis Group), Sizmek, SteelHouse, Storygize and more.
Advertising technology pushes database performance to the limit. Ad-tech platforms must analyze huge amounts of audience data, determine impression value and decide to participate — or not — in the blink of an eye.
“Real-time bidding is an unforgiving business there’s no room for high latency,” said Mina Naguib, Site Reliability Engineering Director at AdGear. “Our platform has to be able to decide and communicate with the exchange in milliseconds if we want to participate in a given bid. With other NoSQL databases, we were unable to take advantage of all bid requests. However, since we migrated to Scylla, it’s no longer an issue.”
To manage the exponential growth of data, many ad-tech companies also need to scale out their databases and optimize the hardware resources they have. Scylla’s distributed, close-to-the-hardware design lets them scale their systems horizontally and vertically while consistently achieving low-single-digit latencies and significantly reducing the number of nodes required for any given workload. Additionally, Scylla recently introduced Scylla Cloud, a fully managed database as a service for organizations looking to reduce administration and capital expenditures.
“Ad-tech companies are at the vanguard of real-time big data, pushing the envelope of scale and performance,” said Dor Laor, CEO of ScyllaDB. “Scylla was purpose-built for high-performance, real-time applications, making it ideal for ad-tech companies that need to minimize latency and maximize the price/performance efficiency of their platforms.”
An open source drop-in replacement for Apache Cassandra, Scylla delivers scale-up performance of millions of OPS per node, scales out to hundreds of nodes and consistently achieves a 99% tail latency of less than 1 millisecond. Scylla’s pioneering shard-per-core implementation, asynchronous architecture and auto-tuning capabilities enable organizations to significantly reduce hardware resources — lowering the total cost of ownership and reducing the complexity of their systems and allow them to meet critical performance metrics.
Numberly’s people-based marketing platform helps brands collect and activate consumer data across their digital channels. After seven years of frustration with MongoDB, Numberly made a move to Scylla.
“We have massive amounts of data coming in from different sources in different formats, which we mix and correlate to help customers engage people with the right message at the right time,” said Alexys Jacob, Chief Technology Officer, Numberly. “It’s critical that we have a database we can trust one that’s not only fast and efficient but maintains consistently low latency. We tried MongoDB and Hive, but neither of them could sustain the kind of workload and low latency our SLAs required. Only Scylla delivered the performance, consistency and stability we needed, and now it’s at the center of everything we do.”
Advertising platform SteelHouse installed its first Scylla cluster in the midst of a major performance crisis. Shortly before Black Friday, with traffic expected to spike up to four times normal volumes, SteelHouse decided to hot-swap its Cassandra database to Scylla. The result was instantly improved stability with far fewer timeouts.
“A lot of people on the business side commented that it was our smoothest holiday season on record. It was also our largest holiday season on record,” said Kevaughn Bullock, Application Architect at SteelHouse. “Had we known beforehand how well Scylla was going to work, we would have obviously made the switch even sooner!”