[dozzie]

From what I know, there is Splunk, which is paid software and I don't know how's its memory requirements, and there's several hosted solutions (Loggly, Papertrail, Sumo Logic). There's also Graylog, but it not only requires ElasticSearch, but also MongoDB, which is worse than just ES.

And pretty much that's it, the whole log aggregation landscape. It bothers me that there's no good open alternative to ElasticSearch, which is only so-so regarding its stability (in operations and in backward compatibility between versions).