Mackerel vs Scandock: monitoring setup, root-cause analysis and onboarding

Mackerel is a Japanese server monitoring service run by Hatena Co., Ltd. You install mackerel-agent on your servers to collect metrics, view them as graphs and raise alerts with monitoring rules. Databases are covered by agent plugins (mackerel-plugin-mysql, mackerel-plugin-postgres and others) or by the AWS integration for Amazon RDS.

Scandock is a service focused on database monitoring. AI reads your tables and data to decide what to watch, and when something is off, it investigates the cause before telling you.

Both help you notice database problems, but they differ in who builds the monitoring and who investigates the cause. In Mackerel, people set the thresholds and people read the graphs to find the cause. In Scandock, AI does both. This page puts screens from both side by side to show the difference.

In short

If you want database monitoring and investigation handled for you, we recommend Scandock.

AI takes over setting and tuning thresholds and comparing graphs to find the cause. You start by registering a connection string, and it also watches anomalies in business data such as orders and payments. If your main goal is watching servers and containers, and you have people who can set thresholds and investigate causes themselves, Mackerel is enough.

Comparison table

Item	Mackerel	Scandock
Designing monitors	You install plugins and create monitoring rules	AI reads your tables and data and decides
Detecting anomalies	You set Warning and Critical thresholds (machine-learning anomaly detection is also available for server system metrics)	AI judges from trends and past outcomes
Investigating the cause	You compare graphs yourself	AI runs queries to investigate and notifies you with evidence
Remediation	You fix things by hand	Registered actions run after you approve them
Setup	Install the agent and plugins on the DB server and edit the config file	Register a connection string
Business data monitoring	Write your own program to post metrics	AI also watches order counts, failed payments and more
Coverage	Servers, containers, cloud, logs, traces and more	Databases only

What to watch: thresholds set by people or by AI

Mackerel

In Mackerel, you choose the metric and the Warning and Critical thresholds for each monitoring rule. You can also set how many points to average and how many consecutive breaches raise an alert. The database plugins collect numbers such as connections, commits and rollbacks, and deadlocks. Deciding how far each number can move before it becomes a problem, and turning that into a threshold, is up to you.

Mackerel also offers anomaly detection for roles, which uses machine learning to spot unusual behavior. It covers system metrics such as CPU and memory collected by mackerel-agent, and does not cover the database plugin metrics (custom metrics).

Mackerel's screen for creating a monitoring rule, with fields for the expression and the Warning and Critical thresholds — Mackerel: creating a monitoring rule (expression monitoring) (source: Mackerel Help, Expression monitoring)

Scandock

In Scandock, AI reads your tables and data and turns things like payment success, failures by payment method, incoming orders and backed-up shipping jobs into monitors. There are no thresholds: it judges anomalies from trends, marks them as fired and moves straight on to investigating the cause.

Finding the cause: reading graphs or receiving an answer

Mackerel

In Mackerel, when an alert arrives, you open the host or role graphs and compare them to see what changed and when. The graphs do not tell you which query or which data caused the problem. Digging into the database is left to you.

Scandock

In Scandock, once AI notices an anomaly, it investigates the database with read-only queries. It summarizes what is happening and where, with evidence, and sends it to Slack or email along with suggested fixes. You read it and decide whether to approve.

Setup steps

Compared for monitoring a self-hosted PostgreSQL database.

Mackerel

Install mackerel-agent on the database server
Install the official plugin package (mackerel-agent-plugins) with yum or apt
Create a monitoring user with the pg_monitor role (PostgreSQL 10 or later)
Add [plugin.metrics.postgres] with the user name and password to mackerel-agent.conf and restart the agent
Create monitoring rules in the UI and set a threshold for each metric

For Amazon RDS, you use the AWS integration instead of the agent. You create an IAM role in AWS, attach read-only policies and register it in Mackerel. It collects CloudWatch metrics, and you still create the monitoring rules yourself.

Scandock

Paste and register a connection string (with an admin user, a read-only monitoring user is created automatically)
AI reads your tables and data, decides on monitors and starts watching

Scandock installs nothing on your database server and does not change database settings. Databases that are not publicly reachable can be connected through Cloudflare Tunnel.

Which one to choose

Mackerel is enough if you

Mainly need to watch servers, containers and cloud resources, and metrics such as connection counts are enough for your databases
Have people who can keep setting and tuning thresholds and investigate the database themselves when an alert fires

Scandock is a good fit if you

Have no database expert, or they are stretched thin
Want to stop setting and tuning thresholds
Want anomalies in business data, such as order counts and failed payments, watched too
Want to spend less time finding the cause after an alert
Want notifications in LINE, Microsoft Teams, kintone and more

In Mackerel, building monitoring rules and finding out what happened inside the database are both work for people. With Scandock, AI designs the monitoring, investigates the cause of an anomaly, notifies you with evidence and runs the fix once you approve it. You can also keep Mackerel for servers and hand database and business data monitoring to Scandock. It is free for 14 days, and it starts watching as soon as you register a connection string.

Sources

Mackerel setup steps and screens were checked on the official site on November 3, 2026. Check the official site for the latest information. Mackerel is a trademark of Hatena Co., Ltd.