The best log management tools for Linux servers in 2026
Grafana Loki is the best choice for most Linux teams: it indexes only labels, so storage stays cheap, and on our test server the single binary ingested and queried logs using about 127 MB of memory. Better Stack is the best hosted option for small teams, with published prices of $0.15 per GB ingested in its US region. Elasticsearch and Kibana remain the most capable for full-text search and analytics, if you can run them or pay for Elastic Cloud.
Vector and Fluent Bit are shippers, not stores: they move logs into one of these tools.
Our top picks
- Grafana LokiBest for cheap, self-hosted logs next to Prometheus and Grafana
- Better StackBest for hosted logs for small teams on a budget
- Elastic Stack (Elasticsearch and Kibana)Best for full-text search and analytics across large log volumes
Compare at a glance
| Tool | Best for | Pricing | Open source | Deployment | Tested by us |
|---|---|---|---|---|---|
| Grafana Loki | cheap, self-hosted logs next to Prometheus and Grafana | Free (self-hosted); in Grafana Cloud, 50 GB of logs a month free, then $0.40 per GB written plus processing and retention on Pro | Yes | Self-hosted, or hosted in Grafana Cloud | Yes |
| Better Stack | hosted logs for small teams on a budget | Free: 10 monitors and heartbeats, 1 status page, 3 GB of logs kept 3 days; Responder seats $29 per month billed yearly ($34 monthly); 50 extra monitors $21 per month yearly; US logs $0.15 per GB ingested plus $0.08 per GB per month retained | No | SaaS | Docs only |
| Elastic Stack (Elasticsearch and Kibana) | full-text search and analytics across large log volumes | Self-managed free tier; Elastic Cloud Serverless Observability from $0.07 per GB ingested and $0.017 per GB per month retained (Logs Essentials); hosted priced by resources | Yes | Self-managed, or Elastic Cloud (hosted or serverless) | Docs only |
| Graylog | a self-hosted log interface with streams and alerts | Graylog Open free; Graylog Enterprise from $15,000 per year and Graylog Security from $18,000 per year (from 10 GB/day) | No | Self-hosted or SaaS (Graylog Cloud) | Docs only |
| OpenSearch | an Apache-2.0 search engine for logs | Free | Yes | Self-hosted | Docs only |
| Datadog Log Management | teams already using Datadog | Ingest $0.10 per GB; Standard indexing with 15-day retention $1.70 per million events billed annually ($2.55 on demand); Flex storage from $0.05 per million events; free trial, no free tier | No | SaaS with the Datadog Agent | Docs only |
| SigNoz | logs, metrics and traces in one open-source app | Community edition free (self-hosted); Teams cloud from $49 per month including $49 of usage, logs and traces $0.30 per GB with 15-day retention; Enterprise from $4,000 per month | Yes | Self-hosted (Community edition) or SigNoz Cloud | Docs only |
| SolarWinds Papertrail | simple hosted log search with S3 archives | From $5 per GB per month; free trial | No | SaaS | Docs only |
| Splunk | large organizations that already standardize on it | Splunk Enterprise and Splunk Cloud Platform: no public price (quote); Splunk Free indexes up to 500 MB a day; Observability Cloud from $15 per host per month billed annually | No | Self-hosted (Splunk Enterprise) or SaaS (Splunk Cloud Platform) | Docs only |
| Vector | shipping and transforming logs into any store | Free | Yes | Self-hosted agent or aggregator | Docs only |
| Fluent Bit | a very small agent on every server or node | Free to use under its open-source license (Apache-2.0) | Yes | Self-hosted agent | Docs only |
The full ranking
Grafana Loki
Best for: cheap, self-hosted logs next to Prometheus and Grafana
Loki is the best fit for most small Linux teams because it keeps storage cheap: it indexes a small set of labels per stream, not the text of every line, and stores compressed chunks on object storage or local disk. You query with LogQL, which works like PromQL, and view results in Grafana. On our AlmaLinux 9.8 test server, the single 3.7.10 binary started from a config of under 30 lines, accepted pushed log lines over HTTP and answered a line-filter query and a count_over_time query, using about 127 MB of memory. Because log text isn't indexed, query speed depends on choosing good labels.
- Pricing
- Free (self-hosted); in Grafana Cloud, 50 GB of logs a month free, then $0.40 per GB written plus processing and retention on Pro (checked November 15, 2026, source)
- License
- Open source (AGPL-3.0)
- Deployment
- Self-hosted, or hosted in Grafana Cloud
- Runs on
- Linux single binary or containers; stores chunks on object storage (such as S3 or GCS) or the local filesystem
Pros
- Cheap to run: indexes labels, not log text
- Single binary for small setups, microservices for large ones
- LogQL and Grafana: the same skills as Prometheus
- Free (AGPL-3.0), or hosted in Grafana Cloud
Cons
- Query speed depends on good label choices
- Needs an agent, such as Grafana Alloy, to ship logs
- No UI of its own; you use Grafana or LogCLI
Tested by us: Loki 3.7.10 on AlmaLinux 9.8, 14 Nov 2026 (UTC): checksum-verified official binary, filesystem storage, bound to localhost; pushed three log lines through the HTTP API and queried them with LogQL (a line filter and count_over_time). About 127 MB of memory. Uninstalled afterwards.
Better Stack
Best for: hosted logs for small teams on a budget
Better Stack's log management is priced openly by volume: in its US region, $0.15 per GB ingested plus $0.08 per GB per month retained, and its European region is cheaper. The free plan, meant for personal projects, keeps 3 GB of logs for 3 days. You get hosted log storage without running a cluster, and the same account covers uptime monitors, status pages and on-call. Its ingest price is close to Datadog's $0.10 per GB, but Datadog also charges per million events to make logs searchable. Longer retention multiplies the storage line, so set it deliberately.
- Pricing
- Free: 10 monitors and heartbeats, 1 status page, 3 GB of logs kept 3 days; Responder seats $29 per month billed yearly ($34 monthly); 50 extra monitors $21 per month yearly; US logs $0.15 per GB ingested plus $0.08 per GB per month retained (checked November 15, 2026, source)
- License
- Proprietary
- Deployment
- SaaS
- Runs on
- Hosted; uptime checks, heartbeats, status pages, on-call and log storage
Pros
- Low per-GB prices, published openly
- Same account for uptime, status pages and on-call
- Free plan to try it
Cons
- Free plan keeps logs only 3 days and is for personal projects
- Prices vary by region
- Hosted only
Not tested by us yet: based on the vendor's documentation.
Elastic Stack (Elasticsearch and Kibana)
Best for: full-text search and analytics across large log volumes
Elasticsearch and Kibana are the reference for full-text log search, aggregations and dashboards. Elastic now offers the source under AGPL-3.0 as one of three license choices, alongside SSPL and the Elastic License 2.0, and the self-managed stack has a free tier. If you'd rather not run the cluster, Elastic Cloud Serverless Observability starts at $0.07 per GB ingested and $0.017 per GB per month retained on Logs Essentials, and hosted deployments are priced by resources. We didn't install it here, because a self-managed cluster is more than our 2 GB test servers should carry. Choose Loki if you mostly filter logs by host and service.
- Pricing
- Self-managed free tier; Elastic Cloud Serverless Observability from $0.07 per GB ingested and $0.017 per GB per month retained (Logs Essentials); hosted priced by resources (checked November 15, 2026, source)
- License
- Source code under AGPL-3.0, SSPL or Elastic License 2.0 (your choice); Elastic's releases under the Elastic License
- Deployment
- Self-managed, or Elastic Cloud (hosted or serverless)
- Runs on
- Self-managed on your own servers, or Elastic Cloud
Pros
- Powerful full-text search, aggregations and Kibana dashboards
- Source available under AGPL-3.0 as one option
- Self-managed free tier, or Elastic Cloud
Cons
- Heavy for small servers; we skipped it on our 2 GB test servers
- Three license options and paid subscription tiers to understand
Not tested by us yet: based on the vendor's documentation.
Graylog
Best for: a self-hosted log interface with streams and alerts
Graylog gives you a purpose-built log interface for search, streams that route messages, alerts and dashboards. Graylog Open is free with no volume limit, but it's source-available under the SSPL rather than an OSI open-source license. A deployment needs Graylog itself, MongoDB for metadata and a Data Node, Graylog's recommended search backend, and the smallest architecture in its documentation uses two servers. Graylog Enterprise starts at $15,000 a year and Graylog Security at $18,000, from 10 GB a day. Pick it if operators need a ready-made log UI; pick Loki if cost and simplicity matter more.
- Pricing
- Graylog Open free; Graylog Enterprise from $15,000 per year and Graylog Security from $18,000 per year (from 10 GB/day) (checked November 15, 2026, source)
- License
- Graylog Open is source-available (SSPL); Enterprise and Security are proprietary
- Deployment
- Self-hosted or SaaS (Graylog Cloud)
- Runs on
- Linux servers (self-managed) or Graylog Cloud
Pros
- Purpose-built search, streams, alerts and dashboards
- Open edition is free with no volume limit
- DEB and RPM packages
Cons
- SSPL, not an OSI open-source license
- Three components to run: Graylog, MongoDB and a Data Node
- Paid editions start at $15,000 a year
Not tested by us yet: based on the vendor's documentation.
OpenSearch
Best for: an Apache-2.0 search engine for logs
OpenSearch is the fork of the last Apache-2.0 versions of Elasticsearch and Kibana, and it's wire-compatible with Elasticsearch 7.10. Everything in the project is under the Apache License 2.0, so there are no license tiers to track, and the license lets you use, modify, embed and resell it. OpenSearch Dashboards covers search and visualization. It brings the same strengths as Elasticsearch and the same weight: a cluster to size, tune and upgrade. Choose it over Elastic when license simplicity matters most; choose Loki if your servers are small.
- Pricing
- Free (checked November 15, 2026, source)
- License
- Open source (Apache-2.0)
- Deployment
- Self-hosted
- Runs on
- Self-hosted clusters on Linux servers or containers
Pros
- Apache-2.0 across the whole project
- Full-text search and OpenSearch Dashboards
- Wire-compatible with Elasticsearch 7.10
Cons
- A cluster to size and operate
- Heavier than Loki for small servers
Not tested by us yet: based on the vendor's documentation.
Datadog Log Management
Best for: teams already using Datadog
Datadog Log Management splits the bill in two. Ingesting costs $0.10 per GB, and indexing the logs you want to search costs $1.70 per million events with 15-day retention (billed annually; $2.55 on demand), while Flex storage for less-searched logs starts at $0.05 per million events. There's a free trial but no free tier. Its strength is context: logs sit next to Datadog metrics, traces and monitors. If you don't already use Datadog, the two-part pricing is harder to forecast than a flat price per GB, because the indexing cost depends on how many events your GBs contain.
- Pricing
- Ingest $0.10 per GB; Standard indexing with 15-day retention $1.70 per million events billed annually ($2.55 on demand); Flex storage from $0.05 per million events; free trial, no free tier (checked November 15, 2026, source)
- License
- Proprietary
- Deployment
- SaaS with the Datadog Agent
- Runs on
- Hosted; collects logs through the Datadog Agent and integrations
Pros
- Logs alongside Datadog metrics, traces and monitors
- Flex storage for cheaper, rarely searched logs
- Ingest everything, index only what you need
Cons
- Two-part pricing (ingest plus indexing) is harder to forecast
- No free tier
- Proprietary
Not tested by us yet: based on the vendor's documentation.
SigNoz
Best for: logs, metrics and traces in one open-source app
SigNoz is an open-source observability platform that puts logs, metrics and traces in one interface. The Community edition is free to self-host with Docker, on Linux, or on Kubernetes with Helm, and most of the code is MIT-licensed, with enterprise directories under a separate license. SigNoz Cloud's Teams plan starts at $49 a month including $49 of usage, with logs at $0.30 per GB and 15 days of retention, and Enterprise starts at $4,000 a month. Choose it if you also want application tracing and prefer one tool to a stack of separate ones.
- Pricing
- Community edition free (self-hosted); Teams cloud from $49 per month including $49 of usage, logs and traces $0.30 per GB with 15-day retention; Enterprise from $4,000 per month (checked November 15, 2026, source)
- License
- Open source core (MIT); enterprise directories under a separate license
- Deployment
- Self-hosted (Community edition) or SigNoz Cloud
- Runs on
- Self-hosted with Docker, on Linux or on Kubernetes (Helm), or SigNoz Cloud
Pros
- Logs, metrics and traces in one UI
- Self-hostable Community edition
- Published cloud prices
Cons
- More to run than a logs-only tool
- Enterprise features are under a separate license
Not tested by us yet: based on the vendor's documentation.
SolarWinds Papertrail
Best for: simple hosted log search with S3 archives
Papertrail, from SolarWinds, is the simplest hosted option here: send it syslog or text log files and search everything in a web log viewer, with team-wide saved searches, search alerts and archives copied to Amazon S3. Pricing starts at $5 per GB per month, with unlimited systems and users and a free trial. Papertrail says it chose its search durations because 3 to 7 days covered about 97% of its own searches, and older logs live in the archives. That design is quick for recent troubleshooting and less convenient for long investigations.
- Pricing
- From $5 per GB per month; free trial (checked November 15, 2026, source)
- License
- Proprietary
- Deployment
- SaaS
- Runs on
- Hosted; accepts logs from servers, VMs, cloud instances, PaaS apps and text log files
Pros
- Very simple setup
- Unlimited systems and users
- Archives to S3 for long-term keeping
Cons
- Higher entry price per GB than usage-based services, though plans bundle search and S3 archives
- Short searchable window; older logs live in archives
- Hosted only
Not tested by us yet: based on the vendor's documentation.
Splunk
Best for: large organizations that already standardize on it
Splunk is an enterprise platform for searching, analyzing and alerting on machine data, and it's rarely the right first choice for a small Linux team. Splunk Enterprise and Splunk Cloud Platform have no public price; you ask for a quote. Splunk Free indexes up to 500 MB a day but has no alerting and no users or roles, so it suits learning, not production monitoring. Observability Cloud, a separate product, starts at $15 per host per month billed annually. Choose Splunk if your organization already runs it; otherwise the tools above cost far less to start.
- Pricing
- Splunk Enterprise and Splunk Cloud Platform: no public price (quote); Splunk Free indexes up to 500 MB a day; Observability Cloud from $15 per host per month billed annually (checked November 15, 2026, source)
- License
- Proprietary
- Deployment
- Self-hosted (Splunk Enterprise) or SaaS (Splunk Cloud Platform)
- Runs on
- Self-hosted or cloud; servers send data through Splunk forwarders
Pros
- Very capable search and analytics
- Self-hosted or cloud
- Free edition for learning (500 MB a day)
Cons
- No public price for Splunk Enterprise or Splunk Cloud Platform
- Splunk Free has no alerting and no user accounts
- Far more than a few servers need
Not tested by us yet: based on the vendor's documentation.
Vector
Best for: shipping and transforming logs into any store
Vector is a shipper, not a store. It collects logs and metrics, transforms them and routes them onward: the project lists 47 sources, 18 transforms and 62 sinks, including Elasticsearch and Amazon S3. It's a single Rust binary that can run as a daemon on each server, as a sidecar or as a central aggregator, and it's open source under MPL-2.0, with Datadog behind the project. Put it in front of any store above, for example to drop noisy lines or parse JSON before you pay to ingest them.
- Pricing
- Free (checked November 15, 2026, source)
- License
- Open source (MPL-2.0)
- Deployment
- Self-hosted agent or aggregator
- Runs on
- Single Rust binary; runs as a daemon, sidecar or aggregator
Pros
- Free and open source (MPL-2.0)
- Transforms can cut volume before you pay for ingest
- Works with many stores (62 sinks)
Cons
- Not a store: you still need one of the tools above
- Another component to deploy and monitor
Not tested by us yet: based on the vendor's documentation.
Fluent Bit
Best for: a very small agent on every server or node
Fluent Bit is also a shipper, not a store. It's a CNCF-graduated agent that collects, processes and forwards logs, metrics and traces through more than 100 built-in plugins, with outputs such as OpenTelemetry, Kafka, Splunk and Amazon S3. The project cites a minimal footprint of about 450 KB, and the agent buffers data to ride out problems at the destination. It's open source under Apache-2.0. Use it when you want the smallest possible agent on each server and a store from this list to send to.
- Pricing
- Free to use under its open-source license (Apache-2.0) (checked November 15, 2026, source)
- License
- Open source (Apache-2.0)
- Deployment
- Self-hosted agent
- Runs on
- Runs as an agent on each server; the project cites a minimal footprint of about 450 KB
Pros
- Very small footprint (about 450 KB minimal, per the project)
- 100+ plugins for inputs, filters and outputs
- CNCF graduated, Apache-2.0
Cons
- Not a store: you still need one of the tools above
- Another component to deploy and monitor
Not tested by us yet: based on the vendor's documentation.
How we picked
We looked for tools that collect logs from Linux servers into one place, keep them as long as you need, and let you search and alert on them. We ranked them on:
- Cost to keep a month of logs, from each vendor's own pricing page, checked on November 15, 2026.
- Search model: full-text indexing, or labels plus filtering.
- Effort to run: components, memory and upgrades for self-hosted tools.
- Alerting on log patterns, and dashboards.
- License and data location: open source, source-available or proprietary; your servers or the vendor's.
We ran Grafana Loki on an AlmaLinux 9.8 test server on November 14, 2026 (UTC), then removed it. We didn't install Elasticsearch, OpenSearch or Graylog, which are too heavy for 2 GB test servers. Everything else is ranked from documentation and pricing pages listed under Sources. Nobody paid to be listed.
Stores and shippers: what's the difference?
A log setup has two halves. A store keeps logs and lets you search them: Loki, Elasticsearch, OpenSearch, Graylog, SigNoz, or a hosted service such as Better Stack, Datadog, Papertrail or Splunk Cloud. A shipper runs on each server, reads logs from the systemd journal and files under /var/log, and sends them to the store: Vector, Fluent Bit, Grafana Alloy or a vendor's own agent. Most stores accept several shippers, so you can change one half without the other. If you only have one server, journalctl may be all you need for now; our journalctl guide covers following and filtering the journal.
What we saw running Loki
We ran the official Loki 3.7.10 binary on a 2 GB AlmaLinux 9.8 server with filesystem storage and a config of under 30 lines, bound to localhost. It reported ready within seconds. We pushed three sample SSH failure lines through the HTTP push API (each returned 204), and then:
- a LogQL query of the form
{job="<our test job>"} |= "Failed password"returned all three lines; count_over_time(... [5m])returned 3, which is how you turn log lines into an alert;- Loki had added a
service_namelabel on its own, alongside ourjobandhostlabels; - the process used about 127 MB of memory.
That's a toy load, but it shows the shape: one binary, one config file and a query language Prometheus users already know.
What should you look for in a log management tool?
Start with volume. Run journalctl --disk-usage and check /var/log on a busy server, multiply by your server count, and you have a rough monthly ingest figure; nearly every price on this page scales with it. Then decide how you'll search. Full-text indexing (Elasticsearch, OpenSearch, Graylog's backend) finds any word fast but costs more storage and memory; label-based stores (Loki) are cheaper but need sensible labels. Check retention: 7 to 30 days covers troubleshooting, while audits may need a year, often in cheaper archive storage. Look for alerts on log patterns, such as a burst of failed logins, not just dashboards. Make sure the shipper understands the systemd journal as well as files. Finally, decide where logs may live, because they often contain IP addresses, usernames and tokens: self-hosting keeps them on your servers, and filtering in the shipper keeps secrets out of any store.
Published log prices, side by side
List prices from each vendor's page on November 15, 2026. They aren't directly comparable, because each vendor splits ingest, indexing and retention differently.
| Service | Ingest | Search and retention | Free allowance |
|---|---|---|---|
| Grafana Cloud (Loki) | $0.40/GB written, plus $0.05/GB processing | $0.10/GB retention; 30 days included on Pro | 50 GB a month, 14 days |
| Better Stack (US region) | $0.15/GB | $0.08/GB per month | 3 GB kept 3 days (personal projects) |
| Elastic Cloud Serverless (Logs Essentials) | From $0.07/GB | From $0.017/GB per month | Trial |
| Datadog | $0.10/GB | $1.70 per million events indexed for 15 days (annual) | Trial |
| SigNoz Cloud | $0.30/GB, 15 days included | Included | $49 a month minimum, including $49 of usage |
| New Relic | $0.40/GB after the free allowance | Not listed separately | 100 GB a month |
| Papertrail | From $5 per GB per month | Search window plus S3 archives | Trial |
| Splunk Free | Up to 500 MB a day | No alerting, no users | Free |
Open source or SaaS?
Self-hosting Loki, OpenSearch, Elasticsearch, Graylog or SigNoz means no per-GB bill and logs that never leave your servers. The cost moves to hardware and time: disks fill, indexes need lifecycle rules, and clusters need upgrades. Loki is the exception that stays small, which is why it ranks first. Hosted services remove that work, and at small volumes they can cost less than the time you'd spend running a cluster, but the bill rises with every noisy log line. Whichever you pick, a shipper such as Vector or Fluent Bit lets you drop debug noise before it reaches a paid store.
Do you need centralized logging at all?
With one server, not always: the systemd journal and journalctl handle searching and following logs well. Central logging earns its keep when you have several servers, when you need logs that outlive a crashed or rebuilt machine, or when an incident needs a single timeline across services. It also helps after a break-in, because logs that live only on the compromised server can be edited by whoever got in.
How we checked this: AlmaLinux 9.8 test server (2 vCPU, 2 GB RAM) on 14 Nov 2026 (UTC): Grafana Loki 3.7.10 from the checksum-verified official binary, with filesystem storage and bound to localhost; we pushed log lines through its HTTP API and queried them with LogQL, then uninstalled it. Elasticsearch, OpenSearch and Graylog were not installed because they are too heavy for our 2 GB test servers. Everything else is researched from vendor documentation and pricing pages checked on 15 Nov 2026. Anything we didn't run ourselves comes from the official documentation listed under Sources.
Check which CVEs are really open on your server
HostCrust Patch Proof compares every installed package with your distribution's own security data. It counts fixes your distro backported and checks the kernel that is actually running, not just the one on disk. Read-only, no root, free.
FAQ
What is the best open-source log management tool?
For most Linux teams, Grafana Loki: it's AGPL-3.0, cheap to store because it indexes only labels, and runs as a single binary for small setups. OpenSearch is the Apache-2.0 choice for full-text search. Graylog Open adds a ready-made log UI but is source-available under the SSPL rather than OSI open source.
Is the ELK stack still free?
Yes, self-managed Elasticsearch and Kibana have a free tier, and since 2024 Elastic offers the source under AGPL-3.0 as one of three license options. Elastic's own releases are under the Elastic License. OpenSearch is the Apache-2.0 fork if you want one simple license.
Should I use Loki or Elasticsearch?
Loki if you mostly find logs by host, service or job and then filter the lines: it's far lighter and cheaper to store. Elasticsearch if you need fast full-text search and heavy analytics across large volumes, and can afford the cluster or Elastic Cloud.
Is Splunk free?
Splunk Free indexes up to 500 MB a day, but it has no alerting and no users or roles. Splunk Enterprise and Splunk Cloud Platform are sold by quote, with no public price.
Are Vector and Fluent Bit log management tools?
They're log shippers. They collect logs on your servers, can parse and filter them, and send them to a store such as Loki, Elasticsearch, OpenSearch or a hosted service. You still need one of those stores to search and keep the logs.
Sources
- Grafana Loki overview: label indexing, storage, LogQL, deployment modes
- Grafana Cloud pricing (logs)
- Grafana Labs licensing (move to AGPLv3)
- Better Stack pricing
- Elastic Cloud Serverless Observability pricing
- Elastic licensing FAQ (AGPLv3, SSPL, Elastic License 2.0)
- Graylog pricing
- Graylog docs: installing Graylog (Data Node, packages)
- Graylog docs: system architecture
- OpenSearch FAQ (Apache 2.0, fork of Elasticsearch 7.10)
- Datadog pricing (Log Management)
- SigNoz pricing
- SigNoz documentation: introduction and install options
- SigNoz on GitHub: free open-source Community edition, MIT outside the ee/ directories
- Papertrail plans
- Splunk pricing
- Splunk docs: about Splunk Free
- Vector
- Vector on GitHub: MPL-2.0 license; maintained by Datadog's Community Open Source Engineering team
- Fluent Bit
- Fluent Bit on GitHub: Apache-2.0 license
- New Relic pricing