John McCalpin

FEBRUARY 17, 2025

The Multicore Era Over the past ~15 years, server processors from Intel and AMD have evolved from the early quad-core processors to the current monsters with over 50 cores per socket. The example below is for a 2005-era processor with 60 ns memory latency and 6.4 If we want to sustain full bandwidth, we need 64/2 =32 cache lines.

Latency 40

Latency Hardware Cache Systems 40

All Things Distributed

JULY 13, 2010

By Werner Vogels on 12 July 2010 05:00 PM. In particular this has been true for applications based on algorithms - often MPI-based - that depend on frequent low-latency communication and/or require significant cross sectional bandwidth. No Server Required - Jekyll & Amazon S3. All Things Distributed. Comments ().

Cloud 96

Cloud AWS Automotive Latency 96

All Things Distributed

MARCH 2, 2011

Japanese companies and consumers have become used to low latency and high-speed networking available between their businesses, residences, and mobile devices. The advanced Asia Pacific network infrastructure also makes the AWS Tokyo Region a viable low-latency option for customers from South Korea. Spot Instances - Increased Control.

AWS 90

AWS Cloud Games Latency 90

All Things Distributed

DECEMBER 3, 2009

This new Region consists of multiple Availability Zones and provides low-latency access to the AWS services from for example the Bay Area. As we announced earlier this month a Region with multiple Availability Zones will come online in Singapore in the first half of 2010, with other regions in Asia to follow later in 2010.

AWS 60

AWS Cloud Latency Storage 60

All Things Distributed

NOVEMBER 19, 2010

By Werner Vogels on 19 November 2010 07:51 AM. for November 2010 was released and an Amazon EC2 Cluster Compute Instance based cluster came in at #231. Understanding Throughput-Oriented Architectures - background article in CACM on massively parallel and throughput vs latency oriented architectures. All Things Distributed.

AWS 62

AWS Cloud Benchmarking Storage 62

All Things Distributed

APRIL 28, 2010

By Werner Vogels on 28 April 2010 11:00 AM. There are four main reasons to do so: Performance - For many applications and services, data access latency to end users is important. The new Singapore Region offers customers in APAC lower-latency access to AWS services. No Server Required - Jekyll & Amazon S3. Comments ().

AWS 60

AWS Cloud Latency Storage 60

All Things Distributed

JULY 11, 2011

As a part of that process, we also realized that there were a number of latency sensitive or location specific use cases like Hadoop, HPC, and testing that would be ideal for Spot. No Server Required - Jekyll & Amazon S3. The Amazon.com 2010 Shareholder Letter Focusses on Technology. Driving down the cost of Big-Data analytics.

AWS 69

AWS Storage Cloud Big Data 69

All Things Distributed

FEBRUARY 24, 2010

By Werner Vogels on 24 February 2010 07:00 AM. Achieving strict consistency can come at a cost in update or read latency, and may result in lower throughput. Lowest read latency. Higher read latency. No Server Required - Jekyll & Amazon S3. The Amazon.com 2010 Shareholder Letter Focusses on Technology.

AWS 63

AWS Latency Database Scalability 63

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. No Server Required - Jekyll & Amazon S3. The Amazon.com 2010 Shareholder Letter Focusses on Technology. Spot Instances - Increased Control.

AWS 88

AWS Government Big Data Cloud 88

Brendan Gregg

JULY 4, 2021

I summarized these topics and more as a plenary conference talk, including my own predictions (as a senior performance engineer) for the future of computing performance, with a focus on back-end servers. Ford, et al., “TCP

Performance 52

Performance Latency Hardware Storage 52

Brendan Gregg

FEBRUARY 28, 2023

My personal opinion is that I don't see a widespread need for more capacity given horizontal scaling and servers that can already exceed 1 Tbyte of DRAM; bandwidth is also helpful, but I'd be concerned about the increased latency for adding a hop to more memory. Ford, et al., “TCP

Performance 52

Performance Latency Cache Virtualization 52

KeyCDN

OCTOBER 24, 2018

With Helsinki being the capital and most populous municipality of Finland, it makes for a great edge server location. Although both countries are relatively close to one another, they are separated by a distance of approximately 500km, which adds up in terms of latency.

Internet 48

Internet Internet Speed Network 48

Alex Russell

MARCH 6, 2021

Sadly, data on latency is harder to get, even from Google's perch, so progress there is somewhat more difficult to judge. We only need to hold the line on script bloat for a few years for devices and networks to overtake the extreme, unconscionable excesses of the 2010's. Time (and chips) can heal these wounds if we only let it.

Performance 127

Performance Network Mobile Metrics 127