Taming Discard Latency Spikes
Learned a crucial lesson on read/write latency when fixing a real ScyllaDB issue! Discover how TRIM requests impact NVMe SSDs with XFS online discard enabled. Uncover the problems and explore potential solutions.
Why Databases Cache, but Caches Go to Disk
ScyllaDB teamed up with Memcached to compare how caches and databases handle storage and memory across different scenarios. We’ll dive into ScyllaDB’s row-based cache vs. Memcached’s in-memory hash table and IO handling.
Primitive Pursuits: Slaying Latency with Low-Level Primitives and Instructions
This talk showcases a methodology with examples to break down applications to low-level primitives and identify optimizations on existing compute instances or platform or for offloading specific portions of the application to accelerators or GPU’s. With the increasing use of a combination of CPU, GPU and accelerators/ASIC’s, this methodology could prove increasingly useful to evaluate what kind of compute to use and when.
How to Improve Your Ability to Solve Complex Performance Problems: Part 2
In Part 2 of my P99 2023 talk, I’ll dive into practical strategies to enhance our problem-solving skills in the performance space. Building on the basics from Part 1, this session will cover actionable tips and methods to improve our odds.
You can watch Part 1 here.
Database Drivers: Performance Perspectives
Unlock the full potential of database drivers! Dive deep into their design, uncover how they work under the hood, and learn to tweak their performance to meet your needs. Perfect for developers eager to optimize their database interactions.
Low-Latency Mesh Services Using Actors
We’re transforming elfo, our Rust actor system, into a distributed mesh of services. Learn how we tackled message serialization, compression, and back-pressure to optimize for high-frequency trading. Dive into our journey to make everything observable and debuggable.
Minimizing Request Latency of Self-Hosted ML Models
Join our session on minimizing latency in self-hosted #ML models in cloud environments. Learn strategies for deploying Deepgram’s speech-to-text models on your hardware, including concurrency limits, auto-scaling, input chunk granularity, and efficient model loading. Optimize your ML inference.
Using Change Point Detection to Fight Noisy Benchmark Results
Discovering performance regressions in modern systems is tough due to inevitable noise. Change Point Detection (CPD) algorithms are gaining traction for tackling this challenge. This talk covers how CPD works and shares examples of real regressions found in open source projects.
Enhancing P99 Latency: Strategies for Doubling/Tripling Performance in Third-Party APIs
Sharing our journey to improve P99 latency in third-party APIs. From optimizing network configs to fine-tuning connection management, we aimed to cut down latency and enhance user experience. Dive into our strategies and see how we achieved a smoother, more responsive service.
Understanding Request Latency with Wallclock Profiling
Analyzing request latency is tough since it’s not always CPU-bound. Many devs give up on CPU profiling, but sampling profilers can help. This talk dives into wallclock profiling and Datadog’s innovative profiling tech to uncover off-CPU activities and unexpected issues.