Datadog vs Mackerel - When to Use a Japan-Made Monitoring SaaS
This article lines up Mackerel, provided by Hatena, against Datadog on three points: the structure of the price list, how hosts are counted, and scope. Using only the structure of the price lists on both official pages, it compares the situations where Mackerel, which you can start with yen-denominated, tax-included pricing, is the lighter choice, and the situations where Datadog pays off when you want to bring the application layer and logs together into one.
Conclusion - Mackerel to Start Host Monitoring Small in Yen, Datadog to Span Layers
Let us state the conditions for choosing between them first. If the monitoring targets are the state of servers and cloud resources (CPU, memory, disk, liveness), the count is from a few to a few dozen hosts, and yen-denominated billing is preferable, Mackerel is the lighter choice. Its price list shows a monthly price per standard host in tax-included yen, and the minimum charge starts at the same amount as one standard host (structure of the price list as of September 2026; Source: Mackerel Pricing). On the other hand, if you want to chase application latency with distributed tracing and cut across logs and real user experience with the same tags, Datadog's structure, where you add products layer by layer, pays off. Host monitoring is Infrastructure Monitoring Pro and Enterprise, priced monthly per host in US dollars, with the unit price varying across the three columns of annual, monthly, and on-demand (as of August 2026; Source: Datadog Pricing). The differences converge on three points: currency and billing form, how hosts are counted, and what lies outside host monitoring, so we lay them out in that order. This article explains how billing works. For current unit prices and plans, refer to each vendor's official pricing page.
Mackerel's Scope and How to Read Its Price List
Mackerel is a SaaS that classifies hosts using the concept of roles and stores the metrics of hosts with mackerel-agent installed at 1-minute intervals (as of September 2026, Source: Mackerel Help Overview). The price list has two plans, Free and Standard, and all Standard unit prices are yen-denominated and tax-included.
| Line item | Billing unit | Counting rule and allotment |
|---|---|---|
| Standard host | Host / month (tax-included JPY) | Hosts with mackerel-agent installed and instances from VM-type integrations. Includes an allotment of host metrics (check the pricing page for the count) |
| Micro host | Host / month (a lower price band than the standard host) | Pods or tasks registered with the container agent as a sidecar, and components from non-VM cloud integrations. Includes a smaller allotment of host metrics than the standard host |
| Additional metric | Metric / month | Posting data every minute for one month counts as 1 metric |
| Trace spans | Per 1 million spans | A free allowance up to a set volume; overage rounded up per 1 million |
| Minimum charge | Per month (the same amount as one standard host) | Months in which a billable feature is used are rounded up to the minimum charge. No charge in months with no use |
The Free plan has caps on the combined host count, metric count, graph display period, monitoring rules, and custom dashboards (check the pricing page for the cap values); in Standard, the graph display period becomes longer and the caps on monitoring rules and dashboards are loosened (some can be relaxed on request). Synthetic monitoring is included in the minimum charge up to a set number of checks, and overage is converted to the equivalent of 1 standard host per bundle of checks. In-role anomaly detection (machine learning) likewise costs the equivalent of 1 standard host per bundle of target hosts, so the design of counting additional features by converting them into host unit prices is applied consistently.
The Counterparts on the Datadog Side Are Infrastructure Pro and Enterprise
What corresponds to Mackerel's standard host in Datadog is Infrastructure Monitoring Pro and Enterprise. The unit price is a monthly price per host in two tiers, Pro and Enterprise, each shown in the three columns of annual, monthly, and on-demand, with the annual commitment the lowest. The Free plan can be used for up to 5 hosts with 1-day metric retention (structure of the price list as of August 2026; Source: Datadog Pricing). Allotments are attached per host: 5 containers and 100 custom metrics with Pro, 10 containers and 200 custom metrics with Enterprise, and metric retention is 15 months for both. Unlike Mackerel, containers and cloud components are counted not as hosts at a separate unit price but as allotments within the host unit price. How to read the USD-denominated price list with its three unit price columns is left to How to Decode the Pricing Structure. Details of the range you can try for free are summarized in Free Plan and Trial.
The Variables to Look at When Comparing the Same Host Count - Currency, Commitment, and Allotments
Because the conclusion would change daily once exchange rates are involved, we do not convert currencies and line up amounts. Under the assumption of "N hosts with the agent installed, no additional containers or cloud components, metrics within the allotment", we lay out the formula that determines the monthly cost and the variables that enter it.
| Configuration | Monthly cost formula | Currency and commitment of the unit price | Examples of included allotments |
|---|---|---|---|
| Mackerel N standard hosts | N x standard host unit price | Tax-included JPY. Monthly credit card charge | Host metric allotment / host, a long graph display period, a set number of synthetic checks |
| Datadog Pro N hosts | N x Pro host unit price | USD. The unit price varies across annual / monthly / on-demand | 5 containers / host, 100 custom metrics / host, 15-month retention |
| Datadog Enterprise N hosts | N x Enterprise host unit price | USD. Same as above | 10 containers / host, 200 custom metrics / host |
What to read in this table is not the absolute amounts but "what is included in the unit price" and "the conditions that move the unit price". Mackerel includes the host metric allotment and a set number of synthetic checks in the base charge, and adds anything beyond that at the metric unit price and in host equivalents. Its unit price does not change with the form of commitment, and being yen-denominated it is not affected by exchange rates. Datadog includes the container and custom metric allotments in the host unit price, and stacks other layers such as APM and logs as separate SKUs. Its unit price changes depending on annual, monthly, or on-demand commitment, and being USD-denominated, the exchange rate becomes a variable against a yen budget. Even with the same host count, the shape of the increase differs: in a container-heavy environment, Mackerel's micro host unit price stacks up per container, while Datadog adds a per-container unit price only for the portion beyond the 5-container allotment. In an estimate, you must always plug in the container count and metric count of your own environment.
Where Mackerel Wins: Billing Form, Low Entry Point, and Japanese-Language Support
The difference starts with the currency of the price list. All of Mackerel's unit prices are yen-denominated and tax-included, payment is a monthly charge to a credit card, and the official FAQ states that receipts are emailed to the owner and can also be obtained from the web console (as of September 2026, Source: Mackerel Pricing). There is no effort of budgeting in yen and no exchange rate difference. The entry point is also low: you can start using paid features from a minimum charge equal to one standard host (tax included), and no charge is incurred in months in which no billable feature is used. On the support side, the same FAQ explicitly states that support on business days from 10:00 to 19:00 is included in the usage fee, and that there is a paid onboarding support program to assist with adoption. The fact that the original text of the help documentation is in Japanese (Source: Mackerel Help Overview) is also a practical benefit as primary information the operations team can read in the original. For a small organization whose targets stay within host monitoring, this difference alone is a reason to choose Mackerel.
Where Datadog Pays Off: Connecting What Lies Outside the Host with the Same Tags
The reason to choose Datadog is determined by what you need outside host monitoring. Its center is a design that lets you conduct an investigation spanning layers, chasing per-request latency in APM, jumping to the logs of the same request in Log Management, and seeing the user-side experience in RUM, on one screen and with one Tag system. Mackerel also has tracing via OpenTelemetry and a log feature (as of September 2026, Source: the FAQ of Mackerel Pricing / Mackerel Help Overview), and spans come with a free allowance up to a set volume, but in Datadog each layer has its own unit price and allotment as an independent product, and the premise is that you add only the layers you need. When you bring in AWS or Azure resources through Integrations, that billing too is counted as a combination of the host unit price allotment and separate SKUs. The condition under which Datadog pays off is "having a plan to use host monitoring as the entry point and extend into latency investigation and log investigation across layers". If host monitoring is all you need, the advantages of this structure are not worth the cost.
Five Points to Look at When Switching
The items to verify when moving from Mackerel to Datadog, or the reverse, are the same in either direction.
- 1. Metric naming: because host metric names and units differ between the two, take inventory on the assumption that every reference name in dashboards and alerts will be rewritten
- 2. Migrating alerts: Mackerel's monitoring rules (capped per plan) and Datadog's Monitors express conditions differently. Put thresholds and notification destinations into a table before rebuilding
- 3. Dashboards: do not assume automatic conversion; build only the few most-viewed ones first, and move the rest only as they become necessary during the parallel period
- 4. Replacing agents: do not swap all hosts at once; switch in stages by role (Mackerel) or tag (Datadog)
- 5. Parallel period: set the period in which both vendors' fees are incurred at 1 to 2 months, and fix the cost ceiling in advance with Mackerel's minimum charge and Datadog's monthly contract
Of the five points, the one directly tied to cost is item 5, the parallel period. Mackerel incurs no charge if no billable feature is used, and Datadog can be stopped the following month on a monthly contract, so the exit from the parallel period can be made light on both sides. Rather than rushing the decision, showing a few hosts to both and recording 2 or 3 times during the parallel period "for the same incident, on which screen did we reach the cause how many minutes faster" gives more reliable material than staring at price lists. For the general procedure for an adoption decision, see the Adoption Checklist, and for migration from OSS, also refer to the comparison with Zabbix.