Usage Scenarios

Loggly makes life simpler for development and DevOps professionals at net-centric companies. Here are the most common usage scenarios adopted by our 6,000+ customers.

Production issue troubleshooting

Your website slows down periodically. Why?

Your customers are complaining about periodic slowness on your site. Google Analytics shows slow page load times but not an increase in page views. Your application servers appear to be healthy. What’s going on?

When you go to Loggly, your saved search for response time shows a spike with timed out queries, on which you zoom in. You search the logs during that time and find a list of warnings for low memory on the production database. You also find the culprit: a slow customer service script lasting several minutes each time it runs.

Before Loggly, you would have to sync and grep through gigabytes of log files, looking at text counts to see patterns in the logs. Now you can see it all on your dashboard in real-time, troubleshooting issues in seconds and putting that extra time toward awesome work.

usage-prodissue-med

usage-alerting-med

Monitoring & alerting

How do I monitor external services?

Your team is responsible for running an order processing web service at a large e-commerce site. It’s big trouble if your customers’ orders aren’t going through.

Your Loggly dashboard shows a bar chart with response times for your service alongside ones for a half dozen dependent services operated by other teams. You know that the payment processing service is a particular area of concern, so you have alerts set up that email when response times frequently fall outside the SLA of 500ms. In addition, the on-call person is paged automatically through PagerDuty.

Before Loggly, order processing would have been down for 30 minutes or more while you figure out which of half dozen services (or your own) has a problem. Now, you can take action before order flow is seriously affected.


Transactional correlation

High-value transactions have not been processed. What’s happening?

A member of your customer support team said that some high-value customer transactions were not going through yesterday. Customers reported they had to exit and try again, and they didn’t offer any specific transaction ID numbers as examples.

In Loggly, you search for transaction logs and then select a filter on the left hand side for events where the transaction status is “initiated” or “completed” and transaction total is over $10,000. You then select the first one where it was initiated but not completed. Searching logs on the transaction ID shows a database update that took 10 minutes to lock up the large amount of inventory, and then rolled back when the user’s window closed. You submit a bug report to fix the performance issue, and tell your support team to remind customers to leave the window open and wait while it processes.

Before Loggly, you would have been relying on guesswork just to start troubleshooting the problem. With Loggly, you are able to take action to fix it within minutes.

usage-transactional-med


usage-trend-med

Trend spotting

Processing costs for MapReduce have risen

Your company’s text analysis engine uses Amazon MapReduce to find and cache affiliate marketing links, but last week your job suddenly required many more hours to complete. Your management team won’t want this news.

You create a graph in Loggly, which shows that processing times have increased and data blocks are balanced across nodes. But a graph of run times shows that one MapReduce task is taking twice as long to run. Looking at the code, you see a bug inserting duplicate entries during the map operation. Not surprisingly, the change was just pushed a week ago. You work with the dev team to patch the bug and test it again. The Loggly dashboard shows now processing times are back to normal, and your EC2 spend is back in line.

Before Loggly, identifying the change among dozens of nodes, tasks, and deployment times could be like finding a needle in a haystack. Loggly’s visual timeline lets you see trends and drill down with a few clicks.


Drill down on alerts from New Relic

Quickly see why a server’s CPU is pegged

You just got an alert from New Relic’s application performance management tool that the CPU is maxed out on one of your servers. You log into New Relic and zoom in on the spike. You then click the Loggly link which automatically searches for the logs from that time interval. In the logs, you see that a daemon is spinning due to an error in the configuration file. Another search shows a recent code change to that server; once you roll it back, the issue disappears.

Before, you would have been moving back and forth between two windows: one for New Relic, and one for Loggly or your terminal window. You would have been copying time ranges and server names for each search. The Loggly New Relic extension for Chrome searches automatically, showing you the relevant logs in a single click.

usage-newrelic-med


usage-analytics-med

Application analytics

Graph your customer activation rate on mobile

Your marketing team wants to know how many new users activate each day, which they define as installing your news reader mobile app and viewing at least one story. Your backend runs on Stackmob with some custom Java code to load the stories and uses syslog4j to send the events to Loggly. With a little bit of help, your marketing team created their own custom dashboard showing the number of new activated users per day. Your app just made the top 10 list in your category last week, and you can see a huge lift!

Before Loggly, you could have used Stackmob’s dashboard to view logs, but showing trends over time would have been a lot of extra work. Mobile analytics platforms like Flurry and Mixpanel require your dev team to insert events into the code and only begin tracking data after your code push. Loggly makes it possible for you to create dashboards from available log data without programming anything.

One-off questions like this one come up all of the time. With Loggly, a huge number of answers are only a few clicks away – because virtually anything can be logged to Loggly.


Integration with other operational systems

Build a custom dashboard for ISVs

Your company works with a variety of independent software vendors (ISVs) to customize and install your system. For customer privacy, you want to give ISVs access to log information for their customers but not all of your customers. You create a custom dashboard that filters log data to just the customer IDs owned by each ISV. Loggly’s RESTful API allows you to search events using a GET request and a list of customer IDs, which returns a list of JSON objects. You even use the open source node-loggly library which makes it even easier to create your dashboard. In fact, you have your first version out in just one day and your boss loves it!

Before Loggly, maintaining any level of privacy for log data would have required setting up a complicated system to aggregate and filter logs in real time. Loggly handles all this for you, with an API that’s easy to use and can process quickly.

usage-integration-med

usage-analytics-med

AWS CloudTrail Support

Monitor modifications to and interactions with your AWS services

Your cloud-based application runs on Amazon Web Services (AWS). An AWS instance just got terminated, but you don’t know who did it. Is your application at risk?

Within Loggly, you have an AWS CloudTrail dashboard, which gives you a visual summary of all modifications to and interactions with your AWS services. In addition, you have alerts set up that email you if an interaction occurs from an unknown IP. In addition, the on-call person is paged automatically through PagerDuty.

Before Loggly, taking action from the audit trail provided through AWS CloudTrail could be a challenge. Now, you know right away that you need to act.

Learn More >>