Brendan Gregg

JULY 4, 2021

## References I've reproduced the talk references below, so you can click on links: - [Gregg 08] Brendan Gregg, “ZFS L2ARC,” [link] Jul 2008 - [Gregg 10] Brendan Gregg, “Visualizations for Performance Analysis (and More),” [link] 2010 - [Greenberg 11] Marc Greenberg, “DDR4: Double the speed, double the latency? Ford, et al., “TCP

Performance 52

Performance Latency Hardware Storage 52

MachMetrics

OCTOBER 16, 2019

However, there is excitement around Starlink for other reasons – namely, the implications it might have for internet speed and latency – even by just a small amount (20 milliseconds on average). Starlink’s Goal: Reduce Internet Latency. What does Starlink and Reduced Latency have to do with me?

Internet 44

Internet Internet Latency Speed 44

Speed Curve

NOVEMBER 3, 2019

” Tammy, Steve and Cliff at Velocity Conference circa 2015. Traditionally, we’ve spent a lot of time focusing on entitled groups as it relates to performance – fast connection speeds, devices with high-end CPUs, low-latency geographies. Building a more inclusive web.

Education 47

Education Latency Strategy Speed 47

Brendan Gregg

MAY 8, 2021

Update: Some modern drives after 2015 are sealed with [helium].) From these outputs I try to determine if the problem is: - **The workload**: High-latency disk I/O is commonly caused by the workload applied. Rotational disks have extra latency from head seeks for random I/O, and spin ups from the idle state. Hit Ctrl-C to end.

Performance 40

Performance Latency Speed Systems 40

Brendan Gregg

JULY 2, 2021

biolatency Disk I/O latency histogram heat map 5. runqlat CPU scheduler latency heat map 10. Here are the top ten tools you can run and present as a generic BPF observability dashboard, along with suggested visualizations: Tool Shows Visualization 1. execsnoop New processes (via exec(2)) table 2. opensnoop Files opened table 3.

Open Source 40

Open Source Latency Cache Energy 40

The Netflix TechBlog

NOVEMBER 22, 2019

December 3 11:30am-12:30pm NFX 208 Netflix’s container journey to bare metal Amazon EC2 Andrew Spyker , Compute Platform Engineering Manager Abstract : In 2015, Netflix started supporting containers as part of their compute platform. We explore all the systems necessary to make and stream content from Netflix.

AWS 37

AWS Entertainment Open Source Benchmarking 37

Adrian Cockcroft

MAY 6, 2023

I don’t advocate “Serverless Only”, and I recommended that if you need sustained high traffic, low latency and higher efficiency, then you should re-implement your rapid prototype as a continuously running autoscaled container, as part of a larger serverless event driven architecture, which is what they did. Finally, what were they building?

Serverless 145

Serverless Lambda Best Practices Traffic 145

All Things Distributed

AUGUST 16, 2011

There are different considerations when deciding where to allocate resources with latency and cost being the two obvious ones, but compliance sometimes plays an important role as well. The move to the cloud is projected by 2015 see a reduction of 30% in IT infrastructure costs, which amounts to $7.2

AWS 111

AWS Government Big Data Cloud 111

Brendan Gregg

SEPTEMBER 25, 2021

A Cassandra database cluster had switched to Ubuntu and noticed write latency increased by over 30%. The change was obvious in the production graphs, showing a drop in write latencies: Once tested more broadly, it showed the write latencies dropped by 43%, delivering slightly better performance than on CentOS.

Speed 126

Speed Java AWS Virtualization 126

Scalegrid

NOVEMBER 25, 2019

Released just four years ago in 2015, Scylla has averaged over 220% year-over-year growth in popularity according to DB-Engines. ScyllaDB offers significantly lower latency which allows you to process a high volume of data with minimal delay. percentile latency is up to 11X better than Cassandra on AWS EC2 bare metal.

Big Data 187

Big Data Database Open Source Azure 187

The Netflix TechBlog

AUGUST 29, 2023

In 2010, however, nearly none of it existed: the CNCF wasn’t formed until 2015! There is a downside to fetching this data on-demand: this adds latency to the first request to a cluster. Today we have a wealth of tools, both OSS and commercial, all designed for cloud-native environments.

Traffic 220

Traffic Latency Cloud C++ 220

All Things Distributed

APRIL 4, 2017

This enables customers to serve content to their end users with low latency, giving them the best application experience. In 2011, AWS opened a Point of Presence (PoP) in Stockholm to enable customers to serve content to their end users with low latency. In 2014 and 2015 respectively, AWS opened offices in Stockholm and Espoo, Finland.

AWS 116

AWS Airlines Latency Games 116

Dynatrace

DECEMBER 22, 2021

Metrics are measures of critical system values, such as CPU utilization or average write latency to persistent storage. But by 2015, it was more common to split up monolithic applications into distributed systems. With observability, teams can understand what part of a system is performing poorly and how to correct the problem.

Artificial Intelligence 180

Artificial Intelligence Metrics Architecture DevOps 180

All Things Distributed

OCTOBER 25, 2018

The new AWS Africa (Cape Town) Region will have three Availability Zones and provide lower latency to end users across Sub-Saharan Africa. In 2015, we expanded our presence in the country, opening an AWS office in Johannesburg. It is our first Region in Africa, and we're shooting to have it ready in the first half of 2020.

AWS 167

AWS Cloud Education Programming 167

Brendan Gregg

SEPTEMBER 25, 2021

A Cassandra database cluster had switched to Ubuntu and noticed write latency increased by over 30%. The change was obvious in the production graphs, showing a drop in write latencies: Once tested more broadly, it showed the write latencies dropped by 43%, delivering slightly better performance than on CentOS.

Speed 40

Speed Java AWS Virtualization 40

Brendan Gregg

MAY 16, 2017

I previously wrote about Netflix in [2015] and [2016], and if you are interested in what it's like to work here, I already covered much in those posts. A latency outlier issue that happened every 15 minutes. I've now worked at Netflix for over three years. Time flies! Java core dump analysis for a crashing JVM. -

Java 75

Java Entertainment Engineering Scalability 75

The Morning Paper

JANUARY 23, 2020

Trade-offs under pressure: heuristics and observations of teams resolving internet service outages , Allspaw, Masters thesis, Lund University 2015. This is part 2 of our look at Allspaw’s 2015 master thesis (here’s part 1 ). 1:06pm reports of the personalised homepage having issues start appearing from multiple sources.

Internet 53

Internet Internet Cache Engineering 53

Brendan Gregg

SEPTEMBER 25, 2021

A Cassandra database cluster had switched to Ubuntu and noticed write latency increased by over 30%. The change was obvious in the production graphs, showing a drop in write latencies: Once tested more broadly, it showed the write latencies dropped by 43%, delivering slightly better performance than on CentOS.

Speed 52

Speed Java AWS Virtualization 52

The Netflix TechBlog

DECEMBER 17, 2019

Passive instances across regions are also possible, though it is recommended to operate in the same region as the database host in order to keep the change capture latencies low. Production usage DBLog is the foundation of the MySQL and PostgreSQL Connectors at Netflix, which are used in Delta. Beresford, and Boerge Svingen.

Database 197

Database Traffic Transportation Open Source 197

The Netflix TechBlog

DECEMBER 17, 2019

Passive instances across regions are also possible, though it is recommended to operate in the same region as the database host in order to keep the change capture latencies low. Production usage DBLog is the foundation of the MySQL and PostgreSQL Connectors at Netflix, which are used in Delta. Beresford, and Boerge Svingen.

Database 212

Database Traffic Transportation Open Source 212

Brendan Gregg

AUGUST 8, 2017

Some people have found values that seem to work for their systems and workloads: they know that when load goes over X, application latency is high and customers start complaining. Latency was acceptable and no one complained. Load averages measured in a modern tool. But there aren't really rules for this. One system with a ratio of 1.5

Latency 111

Latency Metrics C++ Systems 111

Alex Russell

JANUARY 30, 2024

JavaScript-Heavy # Since at least 2015, building JavaScript-first websites has been a predictably terrible idea, yet most of the sites I trace on a daily basis remain mired in script. [1] If you or your company are able to generate a credible worldwide latency estimate in the higher percentiles for next year's update, please get in touch.

Performance 90

Performance Network Mobile Speed 90

Alex Russell

MARCH 6, 2021

Devices and networks have evolved too: Alex Russell @slightlylate An update on mobile CPUs and the Performance Inequality Gap: Mid-tier Android devices (~$300) now get the single-core performance of a 2014 iPhone and the multi-core perf of a 2015 iPhone. The cheapest (high volume) Androids perform like 2012/2013 iPhones, respectively.

Performance 124

Performance Network Cache Metrics 124

Smashing Magazine

AUGUST 9, 2021

You’ve probably heard things like: “HTTP/3 is much faster than HTTP/2 when there is packet loss”, or “HTTP/3 connections have less latency and take less time to set up”, and probably “HTTP/3 can send data more quickly and can send more resources in parallel”. Why Do We Need HTTP/3?

Transportation 145

Transportation Internet Internet Network 145

Smashing Magazine

JANUARY 6, 2020

Estimated Input Latency tells us if we are hitting that threshold, and ideally, it should be below 50ms. Designed for the modern web, it responds to actual congestion, rather than packet loss like TCP does, it is significantly faster , with higher throughput and lower latency — and the algorithm works differently.

Performance 145

Performance Cache Network Metrics 145

Smashing Magazine

JANUARY 7, 2019

Estimated Input Latency tells us if we are hitting that threshold, and ideally, it should be below 50ms. In real-life world, most products aren’t even close: an average bundle size today is around 400KB , which is up 35% compared to late 2015. On a middle-class mobile device, that accounts for 30-35 seconds for Time-To-Interactive.

Performance 103

Performance Cache Metrics Network 103

Smashing Magazine

JANUARY 11, 2021

Estimated Input Latency tells us if we are hitting that threshold, and ideally, it should be below 50ms. Designed for the modern web, it responds to actual congestion, rather than packet loss like TCP does, it is significantly faster , with higher throughput and lower latency — and the algorithm works differently.

Performance 145

Performance Cache Media Metrics 145