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.

Amazon Kinesis

88 pointsby nphaseover 11 years ago

13 comments

zhaodaxiongover 11 years ago
As a team member helped built the service, I would like to offer some of my personal understanding. I am not with Amazon now, and all my views are based on public information on the website.<p>Like all AWS offerings, Kinesis is a platform. It looks like kafka + storm, with fully integrated ecosystem with other AWS services. From the very beginning, the reliability, real-time processing, and transparent elasticity are built in. That&#x27;s all I can say.
mikebabineauover 11 years ago
This is essentially a hosted Kafka (<a href="http://kafka.apache.org/" rel="nofollow">http:&#x2F;&#x2F;kafka.apache.org&#x2F;</a>). Given the complexity of operating a distributed persistent queue, this could be a compelling alternative for AWS-centric environments. (We run a large Kafka cluster on AWS, and it is one of our highest-maintenance services.)
评论 #6735026 未加载
pvnickover 11 years ago
What&#x27;s going on with Amazon recently? We&#x27;re seeing a torrent of new technologies and platform offerings. Are we finally catching a glimpse of Bezos&#x27;s grand scheme?
评论 #6734881 未加载
评论 #6735997 未加载
评论 #6734879 未加载
kylequestover 11 years ago
The 50KB limit on data (base64 encoded data) will be a gotcha you&#x27;ll have to deal with similar to the size limit in DynamoDB. Now you&#x27;ll have to split your messages so they fit inside the Kinesis records and then you&#x27;ll have to reassemble them on the other end... Not fun :-)
kylequestover 11 years ago
Having to base64 encode data is also a bit awkward. They should be passing PutRecord parameters as HTTP headers (which they are already using for other properties) and let users pass raw data in the body.
itchyouchover 11 years ago
It&#x27;s interesting to see these messaging platforms and the new use cases starting to hit the mainstream a la kinesis, storm, kafka.<p>Some interesting things about these kinds of measaging platforms.<p>Many exhanges&#x2F;algo&#x2F;low-latency&#x2F;hft firms have large clusters of these kinds of systems for trading. The open source stuff out there is kind of different from the typical systems that revolve around a central engine&#x2F;sequencer (matching engine).<p>There&#x27;s a large body of knowledge in the financial industry on building low-latency versions of these message processors. Here&#x27;s some interesting possibilities. On an e5-2670 with 7122 solarflare cards running openonload, its possible to pump a decent 2M 100byte messages&#x2F;sec with a packetization of around 200k pps.<p>Avergae latency through a carefully crafted system using efficient data structures and in-memory only stores can pump and process a message through in about 15 microseconds with the 99.9 percent median at around 20 micros. This is a message hitting a host, getting sent to an engine, then back to the host and back.<p>Using regular interrupt based processing and e1000s probably yields around 500k msgs&#x2F;sec with average latency through the system at around 100 micros and 99.9% medians in the 30-40 millisecond range.<p>Its useful to see solarflares tuning guidelines on building uber-efficient memcache boxes that can handle something like 7-8M memcache requests&#x2F;sec.
carterschonwaldover 11 years ago
Before I clicked the link I was hoping Amazon was releasing a clone of the kinesis keyboard. Anyone else have that initial hope? :-)
评论 #6734801 未加载
dylanzover 11 years ago
Can someone with enough knowledge give a high level comparison to Kinesis compared with something like Storm or Kafka?
vosperover 11 years ago
I&#x27;m really excited about this - data streaming has been a crucial missing piece for building large-scale apps on AWS.<p>If the performance and pricing are right it&#x27;s going to relieve a lot of headaches in terms of infrastructure management.
评论 #6734910 未加载
andrewcookeover 11 years ago
<i>it is possible that the MD5 hash of your partition keys isn&#x27;t evenly distributed</i><p>how? i mean, apart from poisson stats &#x2F; shot noise, obviously (and which is noise, so you can&#x27;t predict it anyway).<p>thinking some more, i guess this (splitting and merging partitions in a non-generic way) is to handle when a consumer is slow for some reason. perhaps that partition is backing up because the consumer crashed.<p>but then why not say that, instead of postulating the people are going to have uneven hashes?<p>[edit:] maybe they allow duplicates?
评论 #6734717 未加载
fizxover 11 years ago
Seems like a useful reworking of SQS, but all the hard work is being done in the client: &quot;client library automatically handle complex issues like adapting to changes in stream volume, load-balancing streaming data, coordinating distributed services, and processing data with fault-tolerance.&quot;<p>Unfortunately, there&#x27;s no explanation of the mechanics of coordination and fault tolerance, so the hard part appears to be vaporware.
评论 #6734830 未加载
评论 #6735408 未加载
kylequestover 11 years ago
The Kinesis consumer API is somewhat equivalent to the Simple Consumer API in Kafka. You&#x27;ll have to manage the consumed sequence number yourself. There&#x27;s no higher level consumer API to keep track of the consumed sequence numbers.
评论 #6736526 未加载
kylequestover 11 years ago
Interesting I&#x2F;O limitations in Kinesis:<p>1MB&#x2F;s writes with 1000 writes&#x2F;s 2MB&#x2F;s reads with 5 read&#x2F;s
评论 #6738105 未加载