Brendan Gregg

JULY 2, 2021

biolatency Disk I/O latency histogram heat map. cachestat File system cache statistics line charts. runqlat CPU scheduler latency heat map. For a more recent example, I wrote cachestat(8) while on vacation in 2014 for use on the Netflix cloud, which was a mix of Linux 3.2 execsnoop New processes (via exec(2)) table.

Latency 144

Latency Cache Energy Systems 144

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 103

Performance Latency Cache Virtualization 103

Brendan Gregg

AUGUST 29, 2021

Reads usually have apps waiting on them; writes may not (write-back caching). biolatency From [bcc], this eBPF tool shows a latency histogram of disk I/O. total used free shared buff/cache available Mem: 64414 15421 349 5 48643 48409 Swap: 0 0 0. This is a 64-Gbyte memory system, and 48 Gbytes is in the page cache.

Cache 78

Cache C++ AWS Java 78

Alex Russell

MARCH 6, 2021

A then-representative $200USD device had 4-8 slow (in-order, low-cache) cores, ~2GiB of RAM, and relatively slow MLC NAND flash storage. mid-priced Androids were slightly faster than 2014's iPhone 6. Sadly, data on latency is harder to get, even from Google's perch, so progress there is somewhat more difficult to judge.

Performance 127

Performance Network Metrics Cache 127

ACM Sigarch

MARCH 18, 2019

How many buffers are needed to track pending requests as a function of needed bandwidth and expected latency? Can one both minimize latency and maximize throughput for unscheduled work? These models are useful for insight regarding the basic computer system performance metrics of latency and throughput (bandwidth). Little’s Law.

Systems 60

Systems Latency Performance Analytics 60

ACM Sigarch

MARCH 25, 2019

How many buffers are needed to track pending requests as a function of needed bandwidth and expected latency? Can one both minimize latency and maximize throughput for unscheduled work? The M/M/1 queue will show us a required trade-off among (a) allowing unscheduled task arrivals, (b) minimizing latency, and (c) maximizing throughput.

Systems 53

Systems Latency Performance C++ 53

Abhishek Tiwari

NOVEMBER 3, 2018

Using CDN for the whole website, you can offload most of the website traffic to your CDN which will handle not only large traffic spikes but also reduce the latency of content delivery. They often get blindsided by vendor’s pitch and end-up making decision based on some fancy demos (see my post from 2014 on Adobe AEM ).

Systems 63

Systems Cache Website Network 63