Kafka vs Sumo Logic
When comparing Kafka vs Sumo Logic, the Slant community recommends Kafka for most people. In the question“What are the best log management, aggregation & monitoring tools?” Kafka is ranked 8th while Sumo Logic is ranked 12th. The most important reason people chose Kafka is:
Kafka is optimized for supporting a huge number of users.
Specs
Ranked in these QuestionsQuestion Ranking
Pros
Pro Optimized for performance
Kafka is optimized for supporting a huge number of users.
Pro Native mirroring support
Kafka has native support for mirroring.
Pro Native compression support
Kafka has native support for compression.
Pro Scalable
Sumo logic is entirely cloud based and very scalable.
Pro Flexible licensing model
Licensing cost is primarily determined by daily ingest of logs, however this is averaged out over 30 days instead of locking a user out of their own data after an arbitrary number of license breaches.
Pro Truly multi-tenant
Sumo Logic is truly multi-tenant, a single instance running on the server can serve multiple groups of users.
Pro A large set of supporting Apps
Allows customers to quickly setup and start getting actionable insights from their infrastructure by using Apps that integrate with various different platforms out of the box.
Cons
Con Java is a resource hog
Java is a resource hog, making this far too slow unless you have money to throw at multiple servers with 1/2TB of ram.
Con Need Zookeeper
Kafka can only work with Zookeeper
Con Useless need for collectors
You have to install a plugin on each host to collect logs, the collector is 89MBs and is written in Java. there's no reason to install a Java tool to send syslog data when Linux already does that natively. The memory footprint for Java-based apps is way too high and, in this case, completely unnecessary.
Con Does not support structured data
They don't support RFC5424 standard events
Con Install is very painful
Con Search is very difficult
Here's an example:_sourceCategory=*windows* _sourceName=Security (4771 OR 4768 OR 4776 OR 4625) | parse regex "EventIdentifier = (?<event_id>\d+?);" | parse regex "ComputerName = \"(?<hostname>.+?)\"" | parse regex "(?:Result|Failure|Error) Code:.+?(?<result_code>0x[A-Fa-f\d]+)\b" nodrop | where result_code !="0x0" AND event_id in ("4771", "4768", "4776","4625") | count by hostname
Con Indexing and search are very slow
Sending around 45000 events to it may take more than 3 minutes to show up in the interface.
Once they show up, a search may take up to 32 seconds to return results. On only 45000 events, the search should return in milliseconds.
Con Difficult / Confusing Interface
The service and interface are very confusing.
Con There can be issues with smaller vendors
There may be some issues when using devices and services for smaller vendors which are not officially supported by Sumo Logic.