Why It Matters
Accelerating big‑data pipelines with GPUs dramatically cuts infrastructure spend and latency, allowing Snap to deliver faster experiment results and more responsive product updates. The approach showcases how companies can repurpose idle AI hardware for batch workloads, a strategy increasingly relevant as cloud GPU demand and costs rise.
Key Takeaways
- •Snap cut job costs by 76% using GPU‑accelerated Spark.
- •GPU usage reduced cores 62% and memory footprint 80%.
- •Spark RAPIDS delivered 2‑3× speedups on join‑heavy jobs.
- •Leveraged idle inference GPUs via GKE for batch processing.
- •Zero code changes required; NVIDIA Ether simplified Spark tuning.
Pulse Analysis
Snap processes over 10 petabytes of experiment data each day, a scale that demands rapid, cost‑effective analytics. By integrating NVIDIA Spark RAPIDS into its Google Cloud Dataproc environment, Snap achieved up to three‑fold speed improvements on join‑intensive workloads while maintaining existing PySpark code. The partnership with NVIDIA and Google Cloud enabled a seamless migration—no code rewrites were needed—allowing the engineering platform to meet strict morning SLAs and keep experimentation results fresh for product teams.
The GPU‑accelerated pipelines delivered dramatic efficiency gains: core usage fell 62%, memory footprint dropped 80%, and disk spill vanished by roughly 120 TB. Spark RAPIDS excelled on join‑heavy jobs (≈3× faster) and still provided 1.5‑2× gains on aggregation and union tasks. NVIDIA Ether further simplified Spark configuration across heterogeneous environments, ensuring consistent tuning when workloads fell back from GPUs to CPUs or Dataproc clusters. By repurposing idle inference GPUs on GKE during off‑peak hours, Snap turned otherwise wasted capacity into a powerful batch‑processing resource, with pre‑emptive controls to protect real‑time services.
These results reshaped Snap’s data roadmap, proving that GPU‑first architectures can slash costs while scaling linearly with user growth. The 76% reduction in job expenses and the ability to run daily A/B testing pipelines faster empower engineers to iterate more quickly and deliver richer AR and AI experiences. The three‑way collaboration illustrates a model for other large‑scale social platforms: leverage cloud‑native GPU services, adopt zero‑code acceleration tools, and build flexible scheduling to maximize hardware utilization. Future plans include expanding GPU‑enabled workloads across more teams, further tightening the feedback loop between data insights and product innovation.
Episode Description
Snap processes more than 10 petabytes of experimentation data every single morning—and with NVIDIA GPU-accelerated Apache Spark on Google Cloud, Snap cut job costs by 76%, reduced memory usage by 80%, and eliminated 120 terabytes of disk spill from its pipelines.
Prudhvi Vatala, head of engineering platforms at Snap, joins the NVIDIA AI Podcast to break down how he and his team completely modernized data infrastructure for a social platform serving nearly a billion monthly active users—using NVIDIA cuDF plugin (formerly referred to as NVIDIA RAPIDS plugin) for Apache Spark on Google Kubernetes Engine, with zero application code changes.
🔬Topics covered:
How Snap runs A/B tests at planetary scale using rigorous statistical methods like heterogeneous treatment effect detection and variance reduction
Why Snap reuses idle inference GPUs between 1–5 a.m. for batch data processing—and how it built a Kubernetes-based platform to do it
How NVIDIA cuDF delivered 3x+ speedups on join-heavy Spark jobs with no code rewrites
The full business impact: 76% cost reduction, 62% fewer cores, 80% less memory, 120 TB of spill eliminated
How a three-way partnership between Snap, NVIDIA, and Google Cloud made it possible in just 8–9 months
Chapters:
0:00 Introduction and Snap overview
3:35 What is Snap’s experimentation platform?
4:05 Why experimentation, safety, and privacy are core at Snap
4:52 How A/B testing works at billion-user scale
8:14 Discovering NVIDIA cuDF plugin
9:06 Benchmarking results: join, union, and aggregation jobs
12:00 Reusing idle GPUs overnight via GKE
13:24 Building a bottom-up GPU data platform at Snap
17:48 Results: 76% cost reduction and partnership impact
20:56 Snap’s evolution and what’s next
Learn more:
NVIDIA cuDF: https://developer.nvidia.com/topics/ai/data-science/cuda-x-data-science-libraries/cudf#accel-apache
Comments
Want to join the conversation?
Loading comments...