All Things Distributed

OCTOBER 2, 2017

Performant – DynamoDB consistently delivers single-digit millisecond latencies even as your traffic volume increases. DynamoDB automatically re-distributes your data to healthy servers to ensure there are always multiple replicas of your data without you needing to intervene. Auto Scaling is on by default for all new tables and indexes.

Internet 111

Internet Internet AWS Performance 111

Brendan Gregg

JUNE 30, 2018

sounds like a homework exercise of purely academic value. In some cases, a benchmark may appear to exceed network bandwidth because it returns from a local memory cache instead of the remote target. Once, during a proof of concept, a client reported that latency was unacceptably high for the benchmark: over one second for each request!

Benchmarking 47

Benchmarking Latency Cache Network 47

The Morning Paper

JANUARY 23, 2020

1:18pm a key observation was made that an API call to populate the homepage sidebar saw a huge jump in latency. The process tracing exercise included: Examning IRC transcripts from multiple channels. Members of the team begin diagnosing the issue using the #sysops and #warroom internal IRC channels.

Internet 53

Internet Internet Cache Engineering 53

Brendan Gregg

JUNE 29, 2018

sounds like a homework exercise of purely academic value. In some cases, a benchmark may appear to exceed network bandwidth because it returns from a local memory cache instead of the remote target. Once, during a proof of concept, a client reported that latency was unacceptably high for the benchmark: over one second for each request!

Benchmarking 40

Benchmarking Latency Cache Network 40

The Morning Paper

NOVEMBER 14, 2019

Sharing is caring caching. Taiji’s routing table is a materialized representation of how user traffic at various edge nodes ought to be distributed over available data centers to balance data center utilization and minimize latency. For example, balance utilisation across all data centers, or optimise for network latency.

Traffic 42

Traffic Internet Internet Latency 42

The Morning Paper

DECEMBER 8, 2019

Therefore any programming abstraction must be low latency and the kernel needs to be kept off the path of persistent data access as much as possible. The beauty of persistent memory is that we can use memory layouts for persistent data (with some considerations for volatile caches etc. in front of that memory , as we saw last week).

Hardware 56

Hardware Programming Media Storage 56

John McCalpin

AUGUST 1, 2018

This is an intellectually challenging and labor-intensive exercise, requiring detailed review of the published details of each of the components of the system, and usually requiring significant “detective work” (using customized microbenchmarks, hardware performance counter analysis, and creative thinking) to fill in the gaps.

Hardware 40

Hardware Transportation Performance Latency 40

John McCalpin

AUGUST 1, 2018

This is an intellectually challenging and labor-intensive exercise, requiring detailed review of the published details of each of the components of the system, and usually requiring significant “detective work” (using customized microbenchmarks, hardware performance counter analysis, and creative thinking) to fill in the gaps.

Hardware 40

Hardware Transportation Performance Latency 40