The Netflix TechBlog

SEPTEMBER 24, 2021

Uploading and downloading data always come with a penalty, namely latency. It is worth pointing out that cloud processing is always subject to variable network conditions. The overall ProRes video processing speed is increased from 50GB/Hour to 300GB/Hour.

Cloud 237

Cloud Media Storage Cache 237

Brendan Gregg

JULY 4, 2021

AWS Graviton2); for memory with the arrival of DDR5 and High Bandwidth Memory (HBM) on-processor; for storage including new uses for 3D Xpoint as a 3D NAND accelerator; for networking with the rise of QUIC and eXpress Data Path (XDP); and so on. Ford, et al., “TCP

Performance 140

Performance Latency Hardware Storage 140

All Things Distributed

JUNE 21, 2017

Today, I'm excited to announce the general availability of Amazon DynamoDB Accelerator (DAX) , a fully managed, highly available, in-memory cache that can speed up DynamoDB response times from milliseconds to microseconds, even at millions of requests per second. DynamoDB was the first service at AWS to use SSD storage.

Speed 121

Speed Cache Latency AWS 121

All Things Distributed

OCTOBER 25, 2013

Many of you know Thorsten von Eicken as the founder of Rightscale , the company that has helped numerous organizations find their way onto AWS. The lack of low-latency made that distributed systems (e.g. database replication, fault tolerance protocols) could not benefit from these advances at the network level.

Network 96

Network Latency Operating System Speed 96

High Scalability

SEPTEMBER 8, 2018

RISELabs , those wonderfully innovative folks over at Berkeley, have uplifted their Anna datatabase —a shared-nothing, thread-per-core architecture to achieve lightning-fast speeds by avoiding all coordination mechanisms—to become cloud-aware. No, I don’t think that is because AWS is earning a 355x margin on DynamoDB!

Storage 199

Storage Performance AWS Efficiency 199

Dynatrace

MAY 17, 2023

It is available for the major OS and cloud platforms (for example, Windows, Linux, Solaris, AWS, Azure, and more) and only requires the deployment of a single service to monitor its environment. This allows us to quickly tell whether the network link may be saturated or the processor is running at its limit.

Metrics 176

Metrics Monitoring Database Network 176

Adrian Cockcroft

NOVEMBER 18, 2019

AWS Developer Relations on how the shift from Robot Operating System (ROS) 1 to ROS 2 will change the landscape for all robot lovers. OPN205-R Contributing to the AWS Construct Library (repeats) Using and loving the AWS Cloud Development Kit and want to help make it better? Alejandra Olvera-Novack?—?AWS

Open Source 40

Open Source AWS Lambda Serverless 40

Adrian Cockcroft

NOVEMBER 18, 2019

AWS Developer Relations on how the shift from Robot Operating System (ROS) 1 to ROS 2 will change the landscape for all robot lovers. OPN205-R Contributing to the AWS Construct Library (repeats) Using and loving the AWS Cloud Development Kit and want to help make it better? Alejandra Olvera-Novack?—?AWS

Open Source 40

Open Source AWS Lambda Serverless 40

All Things Distributed

MARCH 2, 2011

Expanding the Cloud - Introducing the AWS Asia Pacific (Tokyo) Region. Today Amazon Web Services is expanding its world-wide coverage with the launch of a new AWS Region located in Tokyo, Japan. The advanced Asia Pacific network infrastructure also makes the AWS Tokyo Region a viable low-latency option for customers from South Korea.

AWS 112

AWS Cloud Games Latency 112

Adrian Cockcroft

JANUARY 20, 2023

Clouds using Ethernet that are multipath optimized using libfabric and features like EFA on AWS are going to be increasingly competitive, and Ethernet will replace other interconnects between racks. AWS has invested in optimizing Ethernet for HPC via the Elastic Fabric Adaptor (EFA) option, and the l ibfabric library.

Architecture 52

Architecture Latency Benchmarking AWS 52

Scalegrid

JANUARY 12, 2024

This is sometimes referred to as using an “over-cloud” model that involves a centrally managed resource pool that spans all parts of a connected global network with internal connections between regional borders, such as two instances in IAD-ORD for NYC-JS webpage DNS routing. This also aids scalability down the line.

Cloud 130

Cloud Virtualization Storage Efficiency 130

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 110

Performance Latency Cache Virtualization 110

Smashing Magazine

AUGUST 3, 2023

Largest Contentful Paint (LCP) LCP measures the perceived load speed of a webpage from a user’s perspective. You need to beware that slow server response times can significantly increase TTFB, often due to server overload, network issues, or un-optimized logic on the server side. Adding to this complexity is the effect of seasonality.

Performance 74

Performance Cache Traffic Metrics 74

Speed Curve

FEBRUARY 5, 2024

Cue server-timing headers Historically, when looking at page speed, we've had the tendency to ignore TTFB when trying to optimize the user experience. The use of server-timing headers by content delivery networks closes a big gap. Latency – How much time does it take to deliver a packet from A to B.

Servers 57

Servers Cache Retail Benchmarking 57

The Morning Paper

FEBRUARY 6, 2020

On the Cloudburst design teams’ wish list: A running function’s ‘hot’ data should be kept physically nearby for low-latency access. Cross-function communication should work at wire speed. A low-latency autoscaling KVS can serve as both global storage and a DHT-like overlay network.

Lambda 98

Lambda Serverless Cache Latency 98

The Morning Paper

MARCH 3, 2020

It takes you through the thinking processes and engineering practices behind the design of a key part of the control plane for AWS Elastic Block Storage (EBS): the Physalia database that stores configuration information. For Physalia, and for AWS more generally, the guiding principle is minimise the blast radius. NSDI’20.

Database 106

Database AWS Network Design 106

All Things Distributed

SEPTEMBER 25, 2017

Today, I am excited to announce plans for Amazon Web Services (AWS) to bring an infrastructure Region to the Middle East! Based in Bahrain, this will be the first Region for AWS in the Middle East. This Region will consist of three Availability Zones at launch, and it will provide even lower latency to users across the Middle East.

Cloud 152

Cloud Education Energy Government 152

Brendan Gregg

JULY 4, 2021

AWS Graviton2); for memory with the arrival of DDR5 and High Bandwidth Memory (HBM) on-processor; for storage including new uses for 3D Xpoint as a 3D NAND accelerator; for networking with the rise of QUIC and eXpress Data Path (XDP); and so on. Ford, et al., “TCP

Performance 52

Performance Latency Hardware Storage 52

The Netflix TechBlog

MARCH 6, 2019

Netflix operates in multiple AWS regions. In addition to improving download speed, this is useful for cutting down on cross-region transfer costs when many workers will be processing the same data? While large buffer sizes speed up sequential data access, they can slow down “sparse” data access?—?that Regional caching? —?Netflix

Media 214

Media Storage Processing Cache 214

All Things Distributed

MAY 13, 2012

With just one click you can enable content to be distributed to the customer with low latency and high-reliability. Of course non-AWS origins are also permitted. This is in addition to the existing optimizations of routing viewers to the edge location with lowest latency for that user, and also persistent connections with the clients.

Cache 117

Cache Latency AWS Website 117

All Things Distributed

OCTOBER 2, 2017

As we began growing the AWS business, we realized that external customers might find our Dynamo database just as useful as we found it within Amazon.com. So, we set out to build a fully hosted AWS database service based upon the original Dynamo design.

Internet 128

Internet Internet AWS Performance 128

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

All Things Distributed

NOVEMBER 29, 2017

Though the AWS Cloud gives you access to the storage and processing power required for ML, the process for building, training, and deploying ML models has unique challenges that often block successful use of this powerful new technology. At AWS, we believe in giving choices, so Amazon SageMaker removes that problem.

Tuning 94

Tuning AWS Scalability Infrastructure 94

O'Reilly

DECEMBER 2, 2019

The most obvious change 5G might bring about isn’t to cell phones but to local networks, whether at home or in the office. High-speed networks through 5G may represent the next generation of cord cutting. Those waits can be significant, even if you’re on a corporate network. What will 5G mean in practice? I don’t, do you?

Wireless 53

Wireless Serverless IoT Artificial Intelligence 53

All Things Distributed

JULY 13, 2010

Customers with complex computational workloads such as tightly coupled, parallel processes, or with applications that are very sensitive to network performance, can now achieve the same high compute and networking performance provided by custom-built infrastructure while benefiting from the elasticity, flexibility and cost advantages of Amazon EC2.

Cloud 118

Cloud AWS Automotive Latency 118

Percona

MAY 2, 2023

The speed of backup also depends on allocated IOPS and type of storage since lots of read/writes would be happening during this process. PBM uses the faster “s2” library and parallelized threads to improve speed and performance if extra threads are available as resources. Records backup progress logging.

Best Practices 52

Best Practices Database Storage Servers 52

Smashing Magazine

JANUARY 11, 2021

Networking, HTTP/2, HTTP/3 OCSP stapling, EV/DV certificates, packaging, IPv6, QUIC, HTTP/3. You need a business stakeholder buy-in, and to get it, you need to establish a case study, or a proof of concept using the Performance API on how speed benefits metrics and Key Performance Indicators ( KPIs ) they care about. Large preview ).

Performance 145

Performance Cache Media Metrics 145

VoltDB

JUNE 1, 2017

It was – like the hypothetical movie I describe above – more than a little bit odd, as you could leave a session discussing ever more abstract layers of virtualization and walk into one where they emphasized the critical importance of pinning a network interface to a specific VM for optimal performance. The “Public Private Cloud” folks.

Hardware 56

Hardware Latency Virtualization Cloud 56

Smashing Magazine

JANUARY 6, 2020

You need a business stakeholder buy-in, and to get it, you need to establish a case study, or a proof of concept using the Performance API on how speed benefits metrics and Key Performance Indicators ( KPIs ) they care about. Start Render time, Speed Index ). Treo Sites provides competitive analysis based on real-world data.

Performance 145

Performance Cache Network Metrics 145

Smashing Magazine

JANUARY 7, 2019

You need a business stakeholder buy-in, and to get it, you need to establish a case study on how speed benefits metrics and Key Performance Indicators ( KPIs ) they care about. Note : If you use Page Speed Insights (no, it isn’t deprecated), you can get CrUX performance data for specific pages instead of just the aggregates.

Performance 104

Performance Cache Network Metrics 104

VoltDB

JUNE 1, 2017

It was – like the hypothetical movie I describe above – more than a little bit odd, as you could leave a session discussing ever more abstract layers of virtualization and walk into one where they emphasized the critical importance of pinning a network interface to a specific VM for optimal performance. The “Public Private Cloud” folks.

Hardware 40

Hardware Latency Virtualization Cloud 40

Brendan Gregg

SEPTEMBER 5, 2017

Here's some output from my zfsdist tool, in bcc/BPF, which measures ZFS latency as a histogram on Linux: # zfsdist. Tracing ZFS operation latency. Both Xen and KVM have had many performance and security improvements, and workloads can now be tuned to run at almost bare metal speeds (say, a 3% loss or less). Hit Ctrl-C to end. ^C

Virtualization 76

Virtualization AWS Engineering Hardware 76

Dynatrace

AUGUST 5, 2019

In this role, I am leading a global team that works closely with our strategic partners such as AWS, Microsoft, Google, Pivotal, Red Hat and others. While most of our cloud & platform partners have their own dependency analysis tooling, most of them focus on basic dependency detection based on network connection analysis between hosts.

Cloud 31

Cloud Traffic Database Network 31