BTW regarding latency: https://engineering.linkedin.com/kafka/benchmarking-apache-kafka-2-million-writes-second-three-cheap-machines
Marko Bonaći Monitoring | Alerting | Anomaly Detection | Centralized Log Management Solr & Elasticsearch Support Sematext <http://sematext.com/> | Contact <http://sematext.com/about/contact.html> On Mon, Sep 5, 2016 at 7:13 PM, Marko Bonaći <[email protected]> wrote: > Hi Gauri, > I'll take a stab at your questions (others will undoubtedly correct me if > I'm wrong). > > Kafka does work over TCP (what else :) > No format expectations (binary). > End to end latency depends on various parameters, like network, memory, > but it is (if you consume straight away; i.e. no consumer lag; i.e. lag can > fit in the OS page cache so you're not even hitting disk when consuming) > measured in low 10s of ms. > No read replicas. You only read from partition's master. I.e. replicas are > used to achieve redundancy. > > > Marko Bonaći > Monitoring | Alerting | Anomaly Detection | Centralized Log Management > Solr & Elasticsearch Support > Sematext <http://sematext.com/> | Contact > <http://sematext.com/about/contact.html> > > On Mon, Sep 5, 2016 at 9:17 AM, Gauri Padbidri <[email protected] > > wrote: > >> Hi All, >> >> >> >> I am very new to Kafka. I am still on the verge of deciding the platform >> to >> choose between Kafka and Kinesis for our requirement. Also it may take me >> more time to come up with a full proof answer, hence thought of >> parallelizing my search. Following are a few queries around Kafka : >> >> >> >> . Does Kafka support sending Streaming Records over raw TCP ? >> >> . Does Kafka expect the packet to be BASE 64 encoded ? >> >> . How much egress and ingress capacity does Kafka support per >> Partition >> ? >> >> . We would be using Kafka for ONLY FOR Data Ingest, do read replicas >> exist for Partitions and if they get replicated to other nodes in the >> Kafka >> cluster? >> >> >> >> Having these queries answered will help me a lot to freeze on Kafka. Look >> forward to your quick response. Thanks ! >> >> >> >> Best Regards, >> >> Gauri Padbidri >> >> >
