Brendan Gregg

SEPTEMBER 25, 2021

A Cassandra database cluster had switched to Ubuntu and noticed write latency increased by over 30%. The broken Java stacks turned out to be beneficial: They helped group together the os::javaTimeMillis() calls which otherwise might have have been scattered on top of different Java code paths, appearing as thin stacks everywhere.

Speed 52

Speed Java AWS Virtualization 52

Brendan Gregg

SEPTEMBER 25, 2021

A Cassandra database cluster had switched to Ubuntu and noticed write latency increased by over 30%. The broken Java stacks turned out to be beneficial: They helped group together the os::javaTimeMillis() calls which otherwise might have have been scattered on top of different Java code paths, appearing as thin stacks everywhere.

Speed 40

Speed Java AWS Virtualization 40

Alex Russell

MAY 9, 2022

This is a complex topic, but to borrow from a recent post , web performance expands access to information and services by reducing latency and variance across interactions in a session, with a particular focus on the tail of the distribution (P75+). Consistent performance matters just as much as low average latency. Who owns this code?

Performance 86

Performance Latency Metrics Engineering 86

Brendan Gregg

OCTOBER 7, 2018

Screenshot: tracing read latency for PID 181: # bpftrace -e 'kprobe:vfs_read /pid == 30153/ { @start[tid] = nsecs; } kretprobe:vfs_read /@start[tid]/ { @ns = hist(nsecs - @start[tid]); delete(@start[tid]); }'. Since I helped developed bpftrace, I'm aware of how fresh my own code is and how likely I introduced bugs. I wrote seeksize.d

C++ 40

C++ Virtualization Programming Latency 40

John McCalpin

JANUARY 22, 2018

Introduction: In December 2017, my colleague Damon McDougall (now at AMD) asked for help in porting the fused multiply-add example code from a Colfax report ( [link] ) to the Xeon Phi x200 (Knights Landing) processors here at TACC. Instead, we found puzzle after puzzle. Instead, we found puzzle after puzzle.

Latency 40

Latency Hardware Code Testing 40

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 Twizzler KVS (key-value store) is just 250 lines of C code, and uses one persistent object for the index structure, and a second one for the data.

Hardware 56

Hardware Programming Media Storage 56

Smashing Magazine

JANUARY 11, 2021

Have we optimized enough with tree-shaking, scope hoisting, code-splitting, and all the fancy loading patterns with intersection observer, progressive hydration, clients hints, HTTP/3, service workers and — oh my — edge workers? It’s much easier to reach performance goals when the code base is fresh or is just being refactored.

Performance 145

Performance Cache Media Metrics 145

HammerDB

OCTOBER 12, 2018

This post complements the previous best practice guides this time with the focus on MySQL and MariaDB and achieving top levels of performance with the HammerDB MySQL TPC-C test. SELECT DISTINCT c from sbtest where id between ? order by c. SELECT c from sbtest where id=? 10 rows in set (0.02

Best Practices 40

Best Practices Scalability Performance C++ 40

Smashing Magazine

JANUARY 7, 2019

Is it worth exploring tree-shaking, scope hoisting, code-splitting, and all the fancy loading patterns with intersection observer, server push, clients hints, HTTP/2, service workers and — oh my — edge workers? Estimated Input Latency tells us if we are hitting that threshold, and ideally, it should be below 50ms.

Performance 103

Performance Cache Metrics Network 103

Smashing Magazine

JANUARY 6, 2020

Is it worth exploring tree-shaking, scope hoisting, code-splitting, and all the fancy loading patterns with intersection observer, server push, clients hints, HTTP/2, service workers and — oh my — edge workers? It’s much easier to reach performance goals when the code base is fresh or is just being refactored.

Performance 145

Performance Cache Network Metrics 145

Brendan Gregg

SEPTEMBER 5, 2017

It uses a Solaris Porting Layer (SPL) to provide a Solaris-kernel interface on Linux, so that unmodified ZFS code can execute. There's also a ZFS send/recv code path that should try to use the TASK_INTERRUPTIBLE flag (as suggested by a coworker), to avoid a kernel hang (can't kill -9 the process). Tracing ZFS operation latency.

Virtualization 76

Virtualization AWS Engineering Hardware 76

Highly Scalable

MAY 1, 2012

This approach often leads to heavyweight high-latency analytical processes and poor applicability to realtime use cases. This process is illustrated in the following code snippet: class LinearCounter { BitSet mask = new BitSet(m) // m is a design parameter void add(value) { int position = hash(value) // map the value to the range 0.m

Analytics 191

Analytics Traffic Big Data Efficiency 191

The Netflix TechBlog

MARCH 13, 2020

SVT-AV1 codebase status The SVT-AV1 repository includes both an AV1 encoder and decoder, which share a significant amount of the code. SVT-AV1 is written in C and can be compiled on major platforms, such as Windows, Linux, and macOS. The SVT-AV1 encoder supports all AV1 tools which contribute to compression efficiency.

Open Source 154

Open Source Efficiency C++ Speed 154

The Netflix TechBlog

JANUARY 24, 2022

Technically, “performance” metrics are those relating to the responsiveness or latency of the app, including start up time. It’s harder to reason about the performance profile of pre-production code since we can’t gather real-time metrics for code that hasn’t yet shipped. and often before they are even committed to the codebase.

Performance 199

Performance Metrics Performance Testing Testing 199

SQL Server According to Bob

JANUARY 11, 2020

Character POS ASCII Value Formula Value A 1 65 67 C 2 67 69 Checksum 136 Comparing ​​ the checksum values indicates ​​ that the values do not match and ​​ damage ​​ has occurred to the data. SQL Server uses read-ahead logic to avoid query stalls caused by I/O waits.

Servers 40

Servers Cache Database Media 40

Smashing Magazine

AUGUST 22, 2021

Because we are dealing with network protocols here, we will mainly look at network aspects, of which two are most important: latency and bandwidth. Latency can be roughly defined as the time it takes to send a packet from point A (say, the client) to point B (the server). Two-way latency is often called round-trip time (RTT).

Performance 105

Performance Network Latency Servers 105

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”. HTTP/2 versus HTTP/3 protocol stack comparison ( Large preview ).

Transportation 145

Transportation Internet Internet Network 145