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.

Gorilla: A Fast, Scalable, In-Memory Time Series Database [pdf]

35 pointsby orrsellaover 9 years ago

8 comments

scurvyover 9 years ago
A few things here:<p>1) What&#x2F;where exactly are they using GlusterFS for? Has Gluster fixed their scaling problems yet? Specifically the issue where new storage spaces&#x2F;nodes were only available to new directories and files, but not existing directories? Granted, the last time I looked at this was 2009 or so, but it was a flaw due to their &quot;no master node&quot; topology.<p>2) FB has an entire team to manage Hadoop&#x2F;HBase. This shows just how much of a beast that stack is. Anyone who has run Hadoop on &quot;Internet time&quot; knows what I&#x27;m talking about. It&#x27;s great at running time insensitive, deferred compute jobs in an academic or scientific setting. It&#x27;s really hard to keep it all 100% running in an on-demand setting. Aside, I couldn&#x27;t imagine just working on 1 product in an operations setting as my full-time job. Boredom&#x2F;fatigue must be a problem on that team.<p>3) I&#x27;d like to see more information on the networking side. What transport protocol? How large are the average updates in frame size? Etc etc.<p>We&#x27;ve built something similar to Gorilla in-house, so I&#x27;m happy to see that we&#x27;ve come to some of the same conclusions.
评论 #10209040 未加载
评论 #10209037 未加载
评论 #10212021 未加载
saosebastiaoover 9 years ago
I really wish this included a comparison with KDB. It&#x27;s not cheap to get a license, and they certainly wouldn&#x27;t give a testing license in order to publish benchmarks against it, but in finance it is the standard for TSDBs. There hasn&#x27;t ever been anything open source that has come close.
评论 #10208863 未加载
nwmcsweenover 9 years ago
Why pointers, why not just do a mirror mmap if you have constant offsets and if time points change and querying based on time points need be constant maybe a table that holds an offset w&#x2F; the difference? Also why not atomics instead of spinning?
评论 #10208747 未加载
rodionosover 9 years ago
&gt; Further, many data sources only store integers into ODS<p>If the underlying data type is 64 bit double, aren&#x27;t they losing precision for integers greater than 2^53?
thrusongover 9 years ago
So this isn&#x27;t managing news feed data or anything like that, it&#x27;s helping them aggregate server performance and error data for quick look up?
rodionosover 9 years ago
Has anyone attended a VLDB conference recently? How is it different from Strata, for example?<p>P.S. Their choice of venues is nice.
pdarshanover 9 years ago
Few folks from Fb started this company called Interana, and they seem to be doing the same thing.
simpsondover 9 years ago
The compression is very neat.