TE
TechEcho
Home24h TopNewestBestAskShowJobs
GitHubTwitter
Home

TechEcho

A tech news platform built with Next.js, providing global tech news and discussions.

GitHubTwitter

Home

HomeNewestBestAskShowJobs

Resources

HackerNews APIOriginal HackerNewsNext.js

© 2025 TechEcho. All rights reserved.

Redis latency spikes and the 99th percentile

168 pointsby r4umover 10 years ago

10 comments

aaronharnlyover 10 years ago
A coworker pointed me to this interesting post noting that in a browser-facing web server (not the same as hits to a datastore), most users will experience the 99th percentile for at least one resource on the page.<p><a href="http://latencytipoftheday.blogspot.it/2014/06/latencytipoftheday-most-page-loads.html" rel="nofollow">http:&#x2F;&#x2F;latencytipoftheday.blogspot.it&#x2F;2014&#x2F;06&#x2F;latencytipofth...</a>
评论 #8533276 未加载
评论 #8533411 未加载
willvarfarover 10 years ago
A nice follow-up and details digging into yesterday&#x27;s &quot;This is why I can&#x27;t have conversations using Twitter&quot; post:<p><a href="https://news.ycombinator.com/item?id=8526208" rel="nofollow">https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=8526208</a>
评论 #8534287 未加载
perlgeekover 10 years ago
Why is forking on xen slow? Most google hits for &quot;xen forking slow&quot; seem to point to some discussions about redis, but I guess other software would suffer from that too.
评论 #8533450 未加载
评论 #8533219 未加载
评论 #8533139 未加载
评论 #8533245 未加载
eamsenover 10 years ago
Side note: you can set you Y axis format to &quot;ms&quot; in Grafana to make the values more descriptive and add the &quot;Points&quot; chart option under display styles to make the mean values visible, which are obscured by the 90th percentile bar in your chart. Also I assume label is wrong, it says 90ths percentile in the chart, but you speak of the 99th percentile.
评论 #8533418 未加载
arielweisbergover 10 years ago
I&#x27;m kind of glad Redis did the fork approach first. It&#x27;s the reason I went with a userspace COW implementation in my work instead of forking and that paid huge dividends. It&#x27;s the difference between starting COW in 10-20 milliseconds versus seconds and most of that time is distributed coordination not flipping the COW boolean.<p>When you crank up the density per node to 256 or 512 gigabytes even bare metal is problematic and in some domains like telecommunications they don&#x27;t care that the spikes are concentrated because they cause cascading failures.<p>I think a userspace COW implementation in Redis would be a big project because you would need a different strategy for every data structure. Being single threaded also makes it challenging to get other&#x2F;more cores to do the serialization and IO. It&#x27;s very doable just not within the current philosophy of thou shalt have one thread per process.
评论 #8534661 未加载
r4umover 10 years ago
fork test by redis labs (2012) <a href="https://redislabs.com/blog/testing-fork-time-on-awsxen-infrastructure#.VFJNvtaiPEI" rel="nofollow">https:&#x2F;&#x2F;redislabs.com&#x2F;blog&#x2F;testing-fork-time-on-awsxen-infra...</a>
评论 #8533236 未加载
toddhover 10 years ago
Curious, why fork in the main thread? Forking traditionally is a pretty heavyweight operation. Perhaps versioning might be more performant?
评论 #8534255 未加载
评论 #8534677 未加载
评论 #8534024 未加载
Beltirasover 10 years ago
I know that redis really wants to be a persistent kvstore. I had a problem with a large website when I increased caching by 2 orders of magnitude (enough RAM to play with). When it came time to write a snapshot to disk, everything died for 5 minutes. Turned it off and haven&#x27;t thought about it since. I&#x27;m not sure I&#x27;ll ever shed my RDBMS predelections.
eldavidoover 10 years ago
This is a huge problem on wall street, where trades must have predictable latencies. Stop-the-world garbage collectors are another source of latency &quot;catastrophes&quot;.
personZover 10 years ago
Does fork not use copy-on-write? I would expect it to add overhead to all overlapped memory operations going forward, but am really surprised if it literally duplicated the entire memory contents.
评论 #8534263 未加载
评论 #8533912 未加载
评论 #8534016 未加载
评论 #8535514 未加载