Remove c
article thumbnail

The Speed of Time

Brendan Gregg

These strange questions came to the fore back in 2014 when Netflix was switching services from CentOS Linux to Ubuntu, and I helped debug several weird performance issues including one I'll describe here. A Cassandra database cluster had switched to Ubuntu and noticed write latency increased by over 30%. How would you _time_ time?

Speed 126
article thumbnail

How To Add eBPF Observability To Your Product

Brendan Gregg

biolatency Disk I/O latency histogram heat map. runqlat CPU scheduler latency heat map. We'll likely open source the new one at some point and have a post about it on the Netflix tech blog. ## Case study: Facebook Facebook are advanced users of BPF, but deep details of how they run the tools fleet-wide aren't fully public.

Latency 145
Insiders

Sign Up for our Newsletter

This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.

Trending Sources

article thumbnail

Analyzing a High Rate of Paging

Brendan Gregg

biolatency From [bcc], this eBPF tool shows a latency histogram of disk I/O. Hit Ctrl-C to end. ^C C msecs : count distribution 0 -> 1 : 83 | | 2 -> 3 : 20 | | 4 -> 7 : 0 | | 8 -> 15 : 41 | | 16 -> 31 : 1620 | * | 32 -> 63 : 8139 | *| 64 -> 127 : 176 | | 128 -> 255 : 95 | | 256 -> 511 : 61 | | 512 -> 1023 : 93 | |. 41 33681 ^C.

Cache 105
article thumbnail

Three Other Models of Computer System Performance: Part 2

ACM Sigarch

With two blog posts, we argue for more use of simple models beyond Amdahl’s Law. Previously, Part 1 of these two blog posts provided our thesis that analytic models can complement measurement and simulations to give quick insight, show what is not possible, provide a double-check, and suggest future directions. and 1/S as ?.

Systems 53
article thumbnail

Three Other Models of Computer System Performance: Part 1

ACM Sigarch

With two blog posts, we argue for more use of simple models beyond Amdahl’s Law. To this end, we present three simple models that we find useful like Amdahl’s Law, split here into two blog posts. How many buffers are needed to track pending requests as a function of needed bandwidth and expected latency? Answered in Part 2.).

Systems 60
article thumbnail

The Speed of Time

Brendan Gregg

These strange questions came to the fore back in 2014 when Netflix was switching services from CentOS Linux to Ubuntu, and I helped debug several weird performance issues including one I'll describe here. A Cassandra database cluster had switched to Ubuntu and noticed write latency increased by over 30%. How would you _time_ time?

Speed 40
article thumbnail

Analyzing a High Rate of Paging

Brendan Gregg

100.00 ^C I'm looking at the r_await column in particular: the average wait time for reads in milliseconds. biolatency From [bcc], this eBPF tool shows a latency histogram of disk I/O. Hit Ctrl-C to end. ^C Hit Ctrl-C to end. ^C avg-cpu: %user %nice %system %iowait %steal %idle 14.81 I think it's the workload. ## 4.

Cache 40