To spot unusual activity on endpoints, including point-of-sale (POS) devices, Sally Beauty’s security team wanted a cleaner way to bring Endgame endpoint detection and response (EDR) data into Elastic Cloud. “Like many data sources, Endgame doesn’t let you pick and choose which components of the data stream you send, and where,” says Sheldon, Information Security Architect for Sally Beauty. That meant some data sent to Elastic Cloud was redundant. Other data wasn’t properly converted to the Elastic Common Schema (ECS), slowing down threat investigation and remediation.
“We wanted a more resilient and flexible data pipeline,” Sheldon says. The immediate need was converting the right Endgame data to ECS, and dropping unneeded fields to make room for additional data sources and retain relevant data for longer.
Intercepting the endgame data stream.
Hearing about Cribl.Cloud, Sheldon conducted an extended evaluation, successfully routing and optimizing the Endgame data stream. “Even better,” he adds, “I carved up the multiple data types in the Endgame data stream, sending each one via its own route to its own pipeline. That means we can route to destinations other than Elastic as needed. With Cribl, we have full control of what data we send, drop, and enhance — and where we send it to take action.”
41% data reduction.
By trimming the Endgame data sent to Elastic Cloud, Sally Beauty gets rid of noise — and frees up space in Elastic to keep data for longer. “I spent a couple of weeks casually working through Cribl pipelines to get the most optimization out of the gate,” Sheldon says. “We’re reducing 9.25TB of daily EDR data down to a little over 5TB a day — 41% reduction. Now we retain data for investigations for 45 days instead of seven.”
“We don’t want our security engineers spending hours becoming wizards in Elastic, Logstash, and syslog-ng. With Cribl, they don’t have to, which frees up more time for our real jobs – security. That’s Cribl’s biggest ROI.”
Sheldon Carmichael
Information Security Architect
Out of the weeds: more time to focus on security.
Cribl frees up compute on the front end and helps the team keep the data for longer retention periods — but the primary win for Sally Beauty is saving time for the security team so they can focus on security — not data engineering. For example, by dropping redundant, blank, or description field clutter, the team can quickly zero in on the fields relevant for investigations. They have the space to bring in backlogged data sources for a more comprehensive view of the company’s security posture. With less data, searches go faster.
Goodnight, syslog-ng server.
Sally Beauty saved even more time by replacing its syslog-ng and Logstash servers with redundant Cribl workers. The cutover was “quick and relatively painless,” according to Carmichael. “Compared to our syslog-ng and Logstash servers, Cribl pipelines are much more straightforward to manage, and they give me both GUI and command line access,” he says. He estimates that his team manages Cribl in one-quarter of the time they spent managing Logstash and syslog. “We don’t want our security engineers spending hours becoming wizards in Elastic, Logstash, and syslog-ng,” says Sheldon. “With Cribl, they don’t have to, which frees up more time for our real jobs — security. That’s Cribl’s biggest ROI.”
The redundant Cribl workers also make the data engine more resilient and easier to manage. The team performs rolling upgrades and maintenance on individual workers without interrupting the data flow.
Tastes like middleware.
Besides moving Endgame data into Elastic, Sally Beauty also uses Cribl to pull IT and Security data from sources that don’t have the capability to push, like their Network Detection and Response (NDR) solution and Microsoft Office365. “Before Cribl, to pull data from our NDR solution and O365, we had to run python scripts on separate servers acting as middleware to pull the data we wanted. That just adds to the systems and resources to maintain, along with everything else we have to manage manually vs in an automated fashion.” says Sheldon. “Cribl allowed us to collapse multiple functions and resources into a single solution with enterprise-grade support,” said Sheldon. This resilience, time savings, and flexibility is what ultimately persuaded Sheldon’s leadership to invest in Cribl.
Looking to the future: using built-in detections.
By trimming the Endgame data sent to Elastic Cloud, Sally Beauty gets rid of noise — and frees up space in Elastic to keep data for longer. “I spent a couple of weeks casually working through Cribl pipelines to get the most optimization out of the gate,” Sheldon says. “We’re reducing 9.25TB of daily EDR data down to a little over 5TB a day — 41% reduction. Now we retain data for investigations for 45 days instead of seven.”
Now Sheldon is planning to use Cribl to modify the Endgame dataset to emulate Elastic Agent data streams. “By doing this we can take advantage of Elastic’s hundreds of built-in detections, and add custom exclusions that persist when Elastic updates its rules, without having to modify each rule every time there’s an upgrade,” he says. The potential time savings are huge. Today the team has to clone the built-in detection rules, modify the observed indices in the rule to include Endgame, and add custom exceptions to tune the rules to Sally Beauty’s environment.
By using Elastic’s built-in detections and the modified data stream from Cribl, Sheldon expects to save significant time supporting upgrades, and to strengthen the company’s security footprint by using more properly tuned and updated detection rules. “A future version of Elastic might give me that capability,” he says. “But the fact that I can do it today with Cribl is pretty awesome, and allows me to apply the same functions to any other observability data we bring into our pipeline.”
Next up for Sally Beauty? Using Cribl to enrich the Endgame stream — for example, by mapping IP addresses to the device’s location (geo-IP) and incorporating threat intelligence. Wrapping up, Sheldon says, “We’ve seen big benefits from Cribl, and I’m a fan. If anyone reading this comes to a Cribl meetup, look me up if you want to talk nerdy.”
TL;DR
Sally Beauty wanted control over which Endgame data it shipped to Elastic Cloud.
Cribl.Cloud parses Endgame data, standardizes it to match the ECS, and reduces it before sending it to Elastic.
Cribl reduced daily EDR data by 41% -- from 9.25TB to 5TB.
The company can now retain EDR data for 45 days instead of 7.
The team replaced syslog-ng and Logstash with Cribl, estimating it takes ¼ of the time to manage Cribl versus the legacy tools.
Eliminated the need to write and manage python scripts to collect “pull” data sources like NDR and O365 logs.
Not having to learn the ins and outs of logging software frees up more time for engineers to work on security.
Selecting Cribl.Cloud gave the team fast time to value and control over cloud latency and egress costs.
