Hi,
does InfluxDB know about data centers?
In a write request I want to make sure that the majority of local nodes got the data, so that a subsequent read is guaranteed to read the newest data.
But I want only local nodes to respond, the other data center should receive the data, but this the other datacenter should not influence the success or timing of my writes (e.g. “local quorum”)
Thanks,
Helge
@HHesse There is a way to setup the system such that this is true. I would use a queueing system (RabbitMQ or other) and have multiple consumers writing to each cluster. Can you share some more about exactly what you are trying to accomplish?
Hi Jack,
my company is processing telemetry data of gas and oil pipelines and I am currently researching a replacement of our time series storage system.
For geo redundancy, we are going to use 2 data centers (on premise).
In each data center, we have a redundant pair of at least 2 nodes (4 nodes total).
We normally write only in one DC, but in case of a network partition between the DC’s, we might write in both.
So we need a mechanism to get eventual consistency between both DC’s.
Do you know about working RabbitMQ/InfluxDB installations?
I would prefere to not care about the other DC, and letting the database do the job.
An alternative to queuing could be to connect all 4 nodes (RF=4)?
Regards,
Helge
@HHesse In this case I would be remiss if I didn’t direct you to the clustered version of our product. We have worked with a number of customers to solve this exact usecase. I can think of at least 3 customers doing cross datacenter replication with RabbitMQ. If you add in Kafka, MQTT, NSQ, etc… that number jumps to the dozens.
