quuxLogging queries across structured and unstructured logs uniformly. While it is best practice to store fully sanitized rich log records, it is exactly when things stop being neat that logs are most critical. If the clue to your emergency is in line 413 of an unformatted stack trace, we still have you covered.
Search anything, anywhere. Use the powerful and well-known syntax of regular expressions to query exactly what you are looking for. Since quuxLogging has no need for log facets, all dimensions on your logs are ready for queries from day one.
Any number in your logs can be aggregated numerically by quuxLogging. Metrics no longer need to be planned and built beforehand and high-dimensionality metrics pose no problem. Combine this with the capability to query unstructured logs and even a time-graph of whatever comes after "bad alloc(), could not allocate" and before "bytes" is trivial.
Percentiles only tell you part of the story. Real systems have bi- or even multi-modal latency distributions, which a percentile plot can not show. A time-series metrics database will let you guess at any correlation between system variables which is not directly time-dependent.
If two numbers are in the same log line, quuxLogging can build a histogram out of that. Response time over response size. Compression ratio over file size. Database latency over query length. Error count over time.
All query options of quuxLogging are available over an API to enable you to build whatever you need beyond our web UI. There will always be a time when you need to do that one weird analysis no one has ever needed before.
We are currently in closed beta and in the process of evaluating a good price point. Our best estimate so far is 0.25 EUR per GB with a full month of retention and 50.00 EUR per PB scanned during queries, reductions due to caching and for off-peak usage apply.
If you log more than 100 GB per month, feel free to contact us at firstname.lastname@example.org.