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.

Ask HN: What are the biggest databases you've worked with?

35 pointsby barelyusableabout 8 years ago
Originally thinking about SQL databases done with PostgresSQL/MySQL, but would be interested in anything. Wondering about number of queries/transactions per second, and how you handled that scale.

9 comments

kevroneabout 8 years ago
At Timehop we currently work with a single instance AWS Aurora (MySQL-ish) database with over 40TB of data (plus a read-only replica on a smaller instance). Some stats: 1.5MB&#x2F;sec receive throughput, 10MB&#x2F;sec transmit throughput, commit latency around 3-4ms (with very occasional spikes to 10-20ms), select &amp; commit counts are about 300&#x2F;s, and select latency hovers around 35ms (we do about a dozen unions per query though).<p>All in all it&#x27;s the easiest relational database I&#x27;ve ever worked with in terms of stability, speed, and scalability. I know this sounds like an ad for Aurora, but I just really like it.
评论 #14250948 未加载
评论 #14256071 未加载
评论 #14245383 未加载
评论 #14250949 未加载
tupshinabout 8 years ago
Having worked with Cassandra for many years, I have worked on:<p>* 1000+ node clusters<p>* Petabyte scale data<p>* 10s of millions of reads and writes per second<p>Given my preface, the &quot;how&quot; is scale out on top of Cassandra, of course. Not SQL, and hard to do if you have a highly relational model, but many stories of success at those kinds of scale.
评论 #14236624 未加载
spthorn60about 8 years ago
MLB&#x27;s Statcast collects 7TB&#x2F;game, or 17 petabytes of raw data annually. <a href="http:&#x2F;&#x2F;fortune.com&#x2F;2015&#x2F;09&#x2F;04&#x2F;mlb-statcast-data&#x2F;" rel="nofollow">http:&#x2F;&#x2F;fortune.com&#x2F;2015&#x2F;09&#x2F;04&#x2F;mlb-statcast-data&#x2F;</a>
thinkMOARabout 8 years ago
Would be nice if people answering include:<p>- hot vs cold data ratio of the total size - read vs write data ratio - if read&#x2F;writes are split - how partitioning, if used, is done - total machine(s) resources disk&#x2F;ram - average (read)query response time - how machine&#x2F;node failure is handled
CyberFonicabout 8 years ago
My biggest installation was an accounting system for a multi-national corporation. 30 Oracle instances running on a 256 core Sun cluster, 192 GB RAM, 40 TB EMC SAN. Typical enterprise system overkill.One of the big 6 consulting firms designed it and deployed PeopleSoft on the completed system. I was just the lowly engineer who configured the hardware, the SAN and the Oracle instances. As for Rolls Royce cars, the performance was &quot;adequate&quot;.
abalashovabout 8 years ago
I&#x27;ve handled ~1 TB DBs in Postgres with about 2000 read queries&#x2F;sec. Technically these were stored function invocations, so wrapped a considerably larger number of queries inside.<p>This didn&#x27;t seem to be a problem. It was the simultaneous write operations that created real limits, banging on the disks&#x2F;disk controller like that.
评论 #14240338 未加载
ohstopituabout 8 years ago
I had worked with ~2 TB of data in CouchDB (a few thousand endpoint calls &#x2F; sec) for my capstone project at University and I thought I had experience, but reading these comments, I realized how much less experience I really have.
avitzurelabout 8 years ago
12TB MongoDB spread across 9 shards (2 replicas per shard) 4TB MySQL with some tables ranging the 400GB size.<p>MongoDB handled about 13K ops&#x2F;sec at peak times with around 5-8K of these being writes.<p>MySQL was probably around 2-3K ops&#x2F;sec.
Clownshoesmsabout 8 years ago
2G, but it was a Cache database, in a hospital. Probably the worst job of my life.