Buyer’s Guide - August 2026

The best ProxySQL monitoring tools, ranked for DBAs

Q: Does ProxySQL have a Prometheus exporter?

Yes, one is built in since ProxySQL 2.1 via the admin REST API. Percona’s standalone proxysql_exporter covers older versions, but it has had no release since October 2021. Note that ProxySQL’s documentation advises against sub-second scraping of the built-in endpoint.

Q: How do I monitor ProxySQL with Grafana?

Enable the built-in exporter, scrape it with Prometheus, and visualize the metrics in Grafana. ProxySQL’s official stack on GitHub ships ProxySQL-Fleet-Overview and ProxySQL-Host-Statistics dashboards to start from. Alternatively, Percona PMM provides a ProxySQL Overview dashboard built on Grafana with the exporter managed for you by pmm-agent.

Q: How do I monitor ProxySQL with Zabbix?

There is no official Zabbix ProxySQL template. Community templates exist, but the most-cited one (zhka/zabbix-proxysql) has had no commits since March 2018. In practice, most Zabbix users build items manually against the stats schema using the MySQL-by-agent technique, since the admin interface speaks the MySQL protocol.

Q: How do I alert when a ProxySQL backend goes down?

Watch for backend status transitions to SHUNNED, OFFLINE_SOFT, or OFFLINE_HARD. Netdata ships ready-made proxysql_backend_shunned, proxysql_backend_offline_hard, and proxysql_hostgroup_no_online_backends alerts. Datadog exposes the proxysql.backend.status service check, which goes CRITICAL on SHUNNED or OFFLINE_HARD. On a DIY Prometheus stack you write these alert rules yourself.

Q: Is ProxySQL monitoring included in Percona PMM?

Yes. PMM is designed to work with ProxySQL and includes a dedicated ProxySQL Overview dashboard. It collects via proxysql_exporter managed by pmm-agent and adds Query Analytics for load-ranked query insight. One gap: ProxySQL-for-PostgreSQL statistics are not covered as of PMM 3.x.

Q: What is the best way to monitor ProxySQL connection pools?

Read stats_mysql_connection_pool per hostgroup and per backend, focusing on ConnFree, ConnUsed, Queries, and Latency. Collection speed matters here: per-second collectors like Netdata catch pool exhaustion faster than 15-second or 60-second scrapes, because pools can fill between samples. Alert on ConnUsed approaching ConnFree and on latency spikes per backend before the pool runs dry.

ProxySQL sits between your applications and MySQL, which makes it the single best place to observe connection pools, query routing, and backend health. It is also the place most monitoring setups forget about. This guide ranks six tools on how completely they read ProxySQL’s stats schema, how fresh the data is, and how much assembly and billing risk comes with each.

Start with Netdata (free) Jump to rankings

The best ProxySQL monitoring tools, ranked for DBAs product interface

Why this list exists

ProxySQL monitoring is its own category, and treating it as just another MySQL instance is the first mistake buyers make. ProxySQL exposes its own stats schema on the admin interface at port 6032: stats_mysql_connection_pool, stats_mysql_global, stats_mysql_query_digest, and stats_mysql_commands_counters. Since version 2.1 it also ships a built-in Prometheus endpoint. A tool that only monitors the MySQL servers behind the proxy, or that only replays ProxySQL’s own monitor module checks (connect, ping, read-only, replication lag), leaves you blind to pool exhaustion, query routing, and backend status transitions.

The second mistake is assuming any metrics pipeline will do. Three dimensions decide the outcome:

Metric depth. Does the tool cover connection pools per hostgroup, query digests, command counters, backend states (ONLINE, SHUNNED, OFFLINE_SOFT, OFFLINE_HARD), query cache, and memory, or just a subset?
Collection resolution. Pool exhaustion and backend latency spikes develop in seconds. Per-second agents catch them; 15-second agent checks and 60-second scrapes often do not.
Alerting and cost shape. Do you get working alerts for SHUNNED backends and hostgroups with no ONLINE backends out of the box, and does the bill grow per node, per host, per GB, or per active series?

One ground rule for this page: we do not quote competitor list prices. Pricing pages change, tiers hide behind sales calls, and a stale dollar figure is worse than none. Instead we describe each vendor’s pricing shape, what actually makes the bill grow, and link the official pricing page so you can check current numbers yourself. For hands-on setup detail, our operator runbooks for ProxySQL walk through the stats schema and the collectors below.

We assembled the shortlist from tools with a documented, maintained path to ProxySQL’s stats schema or its built-in Prometheus endpoint: a native collector, a managed exporter, a packaged Agent integration, or a documented template. Tools whose only answer is “write your own script” were ranked accordingly.

ProxySQL metric depth and collection resolution carry the most weight, because they determine whether you can actually see a pool filling up or a backend being shunned in time to act. Alerting, deployment and cost shape, and setup burden decide whether the tool stays healthy after the proof of concept ends.

Tester credit

Compiled by the Netdata team - Updated August 12, 2026

Sources verified

Scoring criteria

ProxySQL metric depth 25%
Coverage of pools, digests, command counters, backend states, cache, and memory.
Collection resolution 20%
Per-second agents beat 15s checks and 60s scrapes for pool exhaustion.
Alerting and anomaly detection 15%
Built-in backend health alerts versus rules you write yourself.
Deployment and cost shape 15%
SaaS versus self-hosted; per node, per host, per GB, or per series.
Setup and maintenance burden 15%
Packaged integrations versus manual exporter and dashboard assembly.
Ecosystem and extensibility 10%
Grafana, PagerDuty, Slack, and community health around the integration.

Vendor 01 / 06 · #netdata

Netdata

Real-time infrastructure monitoring with a native ProxySQL collector that reads the stats schema every second and ships backend health alerts.

Best for

DBAs and SREs who want per-second visibility into ProxySQL connection pools, backend latency, and query routing without running a separate exporter
Teams that want built-in alerts when backends go SHUNNED or OFFLINE_HARD instead of writing their own rules
Fleet operators who want one pane across many ProxySQL and MySQL nodes with ML-based anomaly detection

Pricing

Per-node pricing: Community plan free up to 5 nodes; Business from $4.50/node/month billed annually, with the per-node price decreasing as node count grows
Unlimited metrics, logs, users, and retention on paid plans; no per-GB or per-metric charges
Agents are open source (GPLv3+) and free; Enterprise on-prem and Homelab plans cover larger and personal fleets

Pros

Native ProxySQL collector (go.d.plugin) reads the stats schema via the admin interface on port 6032 at a default 1-second interval
Covers connection pools, client and server connections, command counters, query cache, jemalloc and internal memory, per-user connection utilization, per-hostgroup backend status, and per-backend latency
Ships three ProxySQL-specific alerts: hostgroup has no ONLINE backends, backend SHUNNED, backend OFFLINE_HARD
Top-queries function surfaces query digest data for spotting hot queries without exporting to another tool
The bill does not grow with metric cardinality, so high-cardinality digest metrics are not a budget risk

Where teams pair it

The ProxySQL collector is not auto-detected; it must be enabled manually, and UI-based configuration requires a paid Netdata Cloud plan
Netdata is observability only: it does not deploy, configure, or fail over ProxySQL the way ClusterControl does
Query insight is a top-queries function rather than PMM-style load-ranked query analytics with explain plans, and alerts on latency percentiles or cache efficiency must be built by the user

Verdict

Netdata leads this list because it is the only tool here with a native ProxySQL collector running at 1-second resolution, built-in alerts for SHUNNED and OFFLINE_HARD backends and for hostgroups with no ONLINE backends, and per-node pricing that ignores metric cardinality. That combination catches pool exhaustion faster than 15-second or scrape-based tools and costs less to predict. The honest caveat: setup is manual, and if you need ProxySQL lifecycle management or deep query analytics you will pair it with ClusterControl or PMM rather than replace them.

Start free ->

Vendor 02 / 06 · #pmm

Percona Monitoring and Management (PMM)

Open-source database observability from Percona with a dedicated ProxySQL Overview dashboard and load-ranked Query Analytics.

Best for

MySQL and Percona shops that want database-specific dashboards plus ProxySQL in one pane
DBAs who want load-ranked query analytics across MySQL and ProxySQL
Teams that prefer open source and can operate their own monitoring server

Pricing

Open source and self-hosted: you run and operate the PMM server and agents, so cost of ownership is your infrastructure and time
PMM Server is AGPLv3 and PMM Client is Apache 2.0
Commercial support subscriptions and hosted options are negotiated with Percona

Pros

Designed to work with ProxySQL, with a dedicated ProxySQL Overview dashboard
Query Analytics ranks queries by load and drills from fleet level down to a single query, deeper query insight than any other tool here
Exporters are managed by pmm-agent, including proxysql_exporter, so there is less manual wiring than a DIY Prometheus stack
Built on Prometheus and Grafana, so existing dashboards and alert rules carry over

Cons

Heavier to operate: you run and maintain the PMM server, agents, and storage yourself
ProxySQL monitoring covers MySQL backends; ProxySQL-for-PostgreSQL statistics are not monitored, per a Percona forum thread
The underlying standalone proxysql_exporter has had no release since late 2021, so new ProxySQL stats depend on PMM’s in-tree maintenance
Prometheus scrape-based collection is coarser than per-second agents

Verdict

PMM is the strongest functional competitor for ProxySQL-specific depth: a dedicated dashboard plus query analytics that Netdata and Datadog do not match. It earns its rank for MySQL-centric shops that already trust Percona tooling. The trade-offs are real, though: you operate the whole stack yourself, granularity is scrape-based rather than per-second, and PostgreSQL-flavored ProxySQL is out of scope.

Visit Percona PMM ->

Vendor 03 / 06 · #prometheus-grafana

Prometheus + Grafana

The open-source metrics stack most ProxySQL deployments pair with, scraping ProxySQL’s built-in Prometheus endpoint and visualizing it in Grafana.

Best for

Teams already standardized on Prometheus and Grafana who want full control of their dashboards
ProxySQL 2.1+ users who can use the built-in exporter instead of maintaining a separate one
Organizations that want everything self-hosted and open source

Pricing

Open source and self-hosted: you operate Prometheus (Apache 2.0) and Grafana (AGPLv3) yourself
Grafana Cloud is usage-based: billed on active series and data points per minute, per-GB logs and traces, and per active user
Cost grows with metric cardinality and retention rather than host count, so high-cardinality query digest metrics can grow the bill

Pros

ProxySQL has shipped a built-in Prometheus exporter since 2.1 (admin-restapi_enabled, default port 6070), removing the need for a separate exporter
ProxySQL’s official Grafana/Prometheus stack provides ProxySQL-Fleet-Overview and ProxySQL-Host-Statistics dashboards
Percona’s proxysql_exporter remains available for older ProxySQL versions and is what PMM uses internally
The Prometheus ecosystem is the de facto standard for metric collection, alerting, and integrations

Cons

Assembly required: you configure the exporter, scrape jobs, retention, dashboards, and alert rules yourself, and keep them current
Percona’s standalone proxysql_exporter has had no release since October 2021, so older-version users rely on stale tooling
Scrape-based collection (the official stack uses 5s) is coarser than per-second agents, and ProxySQL docs advise against sub-second scraping
No ProxySQL-specific alerting out of the box; you write every rule

Verdict

Prometheus and Grafana are the default community answer for ProxySQL metrics, and the built-in exporter since 2.1 removed the worst part of the setup. If your organization already runs this stack well, adding ProxySQL is a scrape job and two imported dashboards. What you give up is resolution, turnkey backend-health alerts, and anyone else’s maintenance burden: the stack is yours to keep alive, and on Grafana Cloud the cardinality bill is yours too.

Read full Netdata vs Prometheus -> Grafana pricing ->

Vendor 04 / 06 · #datadog

Datadog

SaaS observability platform with a ProxySQL integration built into the Datadog Agent that tracks connection pools, memory, and backend status.

Best for

SaaS-first organizations that already run Datadog and want ProxySQL in the same platform
Teams that want turnkey setup: the integration ships inside the Datadog Agent
Teams that want service checks that flag SHUNNED and OFFLINE_HARD backends automatically

Pricing

SaaS, per-host per-month for infrastructure monitoring, with volume discounts
Additional products billed separately: per-GB log ingestion, per custom metric and cardinality, APM per host
The bill grows with host count, log volume, and custom metric cardinality across a fleet

Pros

ProxySQL check is included in the Datadog Agent (integration v7.7.0, minimum Agent 7.19.0) with no separate exporter to run
Collects connection pool, jemalloc and internal memory, command execution-time buckets, query cache, query rule hits, and slow queries
Service checks: proxysql.can_connect and proxysql.backend.status, CRITICAL on SHUNNED or OFFLINE_HARD and WARNING on OFFLINE_SOFT
Can also ingest ProxySQL logs: proxysql.log, query log, and audit log

Cons

SaaS-only: all metrics leave your infrastructure for Datadog’s cloud
Per-host plus per-GB and per-custom-metric billing can grow quickly across a fleet
Default Agent collection interval is 15 seconds, coarser than per-second collectors when pools fill fast
Monitoring and alerting only; no ProxySQL management features

Verdict

Datadog is the turnkey SaaS option: the ProxySQL check ships in the Agent, and the backend-status service check gives you working SHUNNED and OFFLINE_HARD alerting with no rule writing. For teams already paying for Datadog, adding ProxySQL is close to free in effort. The costs are structural: 15-second granularity, SaaS-only deployment, and a bill that scales on three axes at once, which is exactly the kind of growth that surprises teams at renewal.

Read full Netdata vs Datadog -> Datadog pricing ->

Vendor 05 / 06 · #clustercontrol

ClusterControl (Severalnines)

Database lifecycle management platform that monitors ProxySQL through agent-based Prometheus collection and a ProxySQL Overview dashboard.

Best for

Teams that deploy, manage, and monitor MySQL clusters and ProxySQL from one console
Operators who want ProxySQL monitoring alongside cluster deployment, backups, and failover tooling
Organizations that prefer agent-based Prometheus monitoring with dashboards in the UI

Pricing

Open-core: Community edition free, Commercial edition billed per node and billed annually
The bill grows with the number of database and proxy nodes under management
CCX, the SaaS version, is billed hourly

Pros

Agent-based monitoring (SCUMM) with Prometheus since ClusterControl 1.7.0, including a ProxySQL Overview dashboard under Load Balancers
Dashboard covers hostgroup size, client and backend connections, failed connections, routed queries, latency, query cache efficiency, and memory
Combines monitoring with ProxySQL deployment and management, which pure monitoring tools do not offer
Node performance histograms for ProxySQL instances

Cons

Commercial value sits in the paid per-node subscription; the free Community edition is limited
Per-node annual billing makes cost scale with every managed node
Monitoring is Prometheus scrape-based, so granularity is coarser than per-second agents
A heavier platform aimed at cluster lifecycle management; overkill if you only need ProxySQL metrics

Verdict

ClusterControl is the right pick when ProxySQL monitoring is one part of a bigger job: deploying, backing up, and failing over MySQL clusters. No other tool here will manage the proxy for you, and that is genuinely valuable in production. If all you need is metrics and alerts, though, you are buying a lifecycle platform to get a dashboard, at per-node prices and scrape-based resolution.

Visit Severalnines -> ClusterControl pricing ->

Vendor 06 / 06 · #zabbix

Zabbix

Enterprise open-source monitoring platform that can watch ProxySQL, but only via community templates or DIY queries against the stats schema.

Best for

Enterprises already standardized on Zabbix that want ProxySQL in the same platform
Teams comfortable building and maintaining their own templates and queries
Organizations that need a self-hosted, no-license-fee platform at scale

Pricing

Open source and self-hosted: no license fee for the software itself
Paid technical support subscriptions (Silver, Gold, Platinum) and Zabbix Cloud SaaS are optional
Real cost is mostly your own infrastructure and engineering time

Pros

Mature, widely deployed open-source platform with strong alerting and visualization
Can query ProxySQL’s stats schema directly, since it is a MySQL-compatible interface, using Zabbix agent or ODBC items
Community templates exist, such as zhka/zabbix-proxysql (Apache 2.0) with hostgroup and server discovery and query statistics

Cons

No official Zabbix template for ProxySQL; support is community-maintained
The most-cited community template (zhka/zabbix-proxysql) has had no commits since March 2018
Practitioners report struggling to wire ProxySQL into Zabbix (Reddit, November 2024), often ending up building their own solution
Polling-based collection is typically 60 seconds or longer, and there is no ProxySQL-specific dashboard out of the box

Verdict

Zabbix can monitor ProxySQL, but nothing about the path is paved. There is no official template, the best-known community one has been untouched since 2018, and practitioners describe building items by hand against the stats schema. If Zabbix is already your enterprise standard and you have the engineering time, it works. If you are choosing a tool specifically for ProxySQL, every option ranked above it gets you to useful dashboards and alerts with far less DIY.

Read full Netdata vs Zabbix -> Zabbix subscriptions ->

Frequently asked questions

How do I monitor ProxySQL?

ProxySQL exposes its stats schema on the admin interface at port 6032, which you query with the stats user like any MySQL-compatible endpoint. Since version 2.1 it also exposes a built-in Prometheus endpoint via the admin REST API (admin-restapi_enabled, default port 6070). Separately, ProxySQL’s own monitor module tracks backend health with connect, ping, read-only, and replication-lag checks. Tools like Netdata, PMM, Datadog, and Prometheus plus Grafana read these sources; the difference between them is how completely and how often.

What ProxySQL metrics matter most?

Start with the connection pool: free and used connections per hostgroup and per backend from stats_mysql_connection_pool. Then backend status transitions between ONLINE, SHUNNED, OFFLINE_SOFT, and OFFLINE_HARD, since those are your early warning for MySQL trouble. Round it out with per-backend latency, query digests (stats_mysql_query_digest), command counters, query cache efficiency, and memory usage.

Does ProxySQL have a Prometheus exporter?

How do I monitor ProxySQL with Grafana?

How do I monitor ProxySQL with Zabbix?

What is the difference between ProxySQL’s stats schema and monitor schema?

The stats schema holds performance counters: connection pools, query digests, command counters, query rules, and per-user statistics. The monitor schema holds backend health check logs for connectivity, ping, read-only status, and replication lag. Monitoring tools consume both, but they answer different questions: stats tells you what traffic is doing, monitor tells you whether backends are healthy.

How do I alert when a ProxySQL backend goes down?

Is ProxySQL monitoring included in Percona PMM?

What is the best way to monitor ProxySQL connection pools?

The best ProxySQL monitoring tools, ranked for DBAs

How we evaluated ProxySQL monitoring tools

Netdata

Percona Monitoring and Management (PMM)

Prometheus + Grafana

Datadog

ClusterControl (Severalnines)

Zabbix

Frequently asked questions

Book Your Free Demo