As I've been building out our Splunk installation I've been treating the indexers as appliances. By that I mean all of the indexers are identical down to every indexer having every index. This allows me to have all forwarders writing to all of the indexers.
My thinking goes something like this:
So, this raises some questions:
Thanks!
The answer is: "it depends". There is a spectrum of deployments out there from one laptop to scores of indexers. Beyond data volume, you should also analyze your search use cases.
Scaling horizontally and treating indexers as disposable commidities is generally the right approach. More, cheaper indexers is generally better than fewer, more expensive indexers. The main cost is management/deployment complexity.
For more best practices, see:
The answer is: "it depends". There is a spectrum of deployments out there from one laptop to scores of indexers. Beyond data volume, you should also analyze your search use cases.
Scaling horizontally and treating indexers as disposable commidities is generally the right approach. More, cheaper indexers is generally better than fewer, more expensive indexers. The main cost is management/deployment complexity.
For more best practices, see: