What do you like best?
I like that with Loggly and some software(Winston, Fluentd) I can capture, interrogate and analyze my application's log output across a variety of deployment platforms and arrangements.
I can log both my development and live servers to the same system and either join or separate their content at will. The existing collection tools for Loggly are very strong and the setup for those tools is incredibly easy. Beyond that they offer the necessary endpoints and formats to receive logs from a wide variety of candidates.
Everything from system logs on various servers to application and routing logs from custom solutions is available.
Loggly essentially runs an Elastic Stack on behalf of the user, and having set up some similar solutions to test, I can declare with confidence that the time and energy saved by *not* rolling my own logging solution(and thereby having to maintain my maintenance tools) is beyond measure.
Support is quick and helpful within reason, I've had nothing resembling a technical difficult caused by failure on Loggly's part.
What do you dislike?
Cos:, While I think the price is reasonable for the amount of work saved, and at $99 a month it pays for itself immediately, placing things like the LiveTail feature(emulating tail -f) at a $250 a month price point is disappointing. Features like that are excellent at every level and I'm simply never gonna convince my superiors to spend more on our log solution than on the live servers it monitors.
Finicky: It's a pain to get your log data arranged so that the various features of the system come into play correctly. I still periodically get unparsed log entries because 2 consecutive log entries had different object signatures for the same field(including single objects vs arrays).
Query Language: There are a few ways to filter/limit data for display and analysis with Loggly, but sometimes things don't work like you think they will and it's not always clear why. Somtimes a query for field:'2' finds objects with field=2, but sometimes it finds nothing whatsoever. When this happens there's always doubt as to whether you structured your query wrong or if there's literally no data meeting the criteria to be found.
Recommendations to others considering the product
Look at their integrations, determine if you can use their tools to build the sort of structure you need.
Spend some real dev time reducing log noise(we ran a search for "console.log" and removed all of them, then went back and started adding calls to our logger explicitly and carefully) and building in the necessary components to standardize the log entry process.
Look into Fluentd, it's excellent for routing, filtering and translating logs, it can help as an adapter to make sure Loggly gets what it needs from where it needs.
What business problems are you solving with the product? What benefits have you realized?
Logs for server-side code are a right pain. Parsing and analyzing logs from a cluster of containers all outputting logs directly to console.log is a nightmare.
Simply put Loggly offers a robust system with simple inputs that allows you to funnel log data to a common, web-accessible location where it can be stored and interacted with in a number of robust ways. Before Loggly our hands were often tied if a user complaint fell into server code, there was simply not an easy way to parse logs from several servers, some of which may have been drained and terminated, and meaningfully determine what was going on.
Loggly gave me the ability to search and sort log data to find the source of problems.
Within a week of completing our integration of Loggly we were discovering the source and profile of bugs that had eluded us for 6 months.
With further adaptation of our web app to the new logging approach, we can start to see traffic patterns and analyse user behavior from the logs as well.
Lastly, Loggly prevents us from taking on a new maintenance and development project in order to analyze our main maintenance and development project. Loggly lets us break the potentially infinite chain of working on tools to work on tools.