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.

Time Structured Merge Tree: From LSM Tree to B+Tree and Back Again

97 pointsby pauldixover 9 years ago

5 comments

gtrubetskoyover 9 years ago
IMHO if your volume is not in 300K data points per second, there is still plain old PostgreSQL, which is perfectly suitable for storing time-series and is time-tested and very reliable, allowing you to keep your TS data alongside your other (business) data so that you can analyze it all together without having to extract out of unfamiliar storage. I&#x27;ve hacked a project which demonstrates how it can be done, see <a href="https:&#x2F;&#x2F;github.com&#x2F;grisha&#x2F;timeriver" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;grisha&#x2F;timeriver</a> and my blog explaining the technique <a href="http:&#x2F;&#x2F;grisha.org&#x2F;blog&#x2F;2015&#x2F;09&#x2F;23&#x2F;storing-time-series-in-postgresql-efficiently&#x2F;" rel="nofollow">http:&#x2F;&#x2F;grisha.org&#x2F;blog&#x2F;2015&#x2F;09&#x2F;23&#x2F;storing-time-series-in-pos...</a>
评论 #10348799 未加载
评论 #10347564 未加载
评论 #10348008 未加载
评论 #10351389 未加载
评论 #10347011 未加载
kbodyover 9 years ago
So, we were using InfluxDB on v0.8.x but had two data corruption incidents. Both times I tried the recovery utilities with no luck. Even with some data loss, it was time to move to another db.<p>We were only getting ~1k tps with some spikes of 10k tps. I hope for the rest of the users, there was some work done of that front. Maybe in the future we will revisit out of curiosity, since we have no issues at the moment.
评论 #10346642 未加载
skybrianover 9 years ago
I&#x27;m wondering if a new low-level db will come out of this as an alternative to LevelDB, BoltDB, and so on? Or do you have to use InfluxDB?
评论 #10350183 未加载
shubhra51over 9 years ago
Woot! Nice throughput improvement over traditional LSM types with both read and write accelerated.
philsnowover 9 years ago
typo in the article&#x27;s headline: Strucutred -&gt; Structured