Datadog vs IBM Instana - How It Differs from an Auto-Instrumenting APM
IBM Instana is an APM of the "install the agent and instrumentation comes with it" type, and its price is determined solely by MVS (hosts). Datadog combines library instrumentation with a per-host price and span allotments. Working from the structure of both official pricing pages, this article lines up the differences in billing units and works through the variables that decide which comes out ahead and the points to verify when switching. For current unit prices, refer to each vendor's official page.
Conclusion - The Philosophical Difference Is "Instrumentation That Comes with Installation" vs "Instrumentation with Allotments"
Both are APMs billed by host count, but what is inside the per-host price differs. Instana places "Automatic discovery and monitoring" among the core capabilities in its official documentation, and is designed to discover and instrument automatically through the agent, sensors, and AutoTrace (as of September 2026, Source: Instana SaaS documentation). The footnote on the pricing page states that the amount paid is the number of MVS (Managed Virtual Servers) multiplied by the unit price (same date, Source: Instana pricing page). Datadog's APM assumes library instrumentation, and on top of the per-host price there are allotments for span ingestion volume and indexed span count, with overages billed separately.
| Aspect | IBM Instana | Datadog APM |
|---|---|---|
| How instrumentation is applied | Automatic discovery via host agent + sensors + AutoTrace | Library instrumentation via SDK or OpenTelemetry |
| Unit of pricing | MVS (physical machine / virtual server / worker node) | Host (used together with the container allotment of Infrastructure Monitoring) |
| Variables outside the unit price | None (amount paid = MVS count x unit price) | Overage beyond the allotments for ingested span GB and indexed span count |
| Contract forms | Annual contract (with a minimum order quantity), hourly-billed PayPerUse, Self-Hosted | Annual contract / monthly contract / on-demand |
What should be compared is not how high or low the unit price is but "what falls outside the unit price": Instana has little that falls outside, while Datadog is a structure in which you buy the breadth to tie everything outside APM (logs / RUM / infrastructure) together with the same tags.
How to Read Instana's Price List - The Unit Called MVS
Instana's official pricing page (as of September 2026) lists three plans - PayPerUse (pay-as-you-go per MVS hour), SaaS (a monthly amount per MVS), and Self-Hosted (an annual amount per MVS) - each shown as a "from" lower-bound value (Source: Instana pricing page). The Japanese-language page shows the same three plans in yen, excluding tax (same date, Source: Japanese-language pricing page). An MVS counts a physical / virtual server or a worker node as 1, and in SaaS you choose either Essentials (infrastructure) or Standard (full stack) for each MVS. The unit price for Standard on its own is not displayed, and the footnote says that prices are indicative, vary by country, exclude tax, and that the amount paid is the MVS count multiplied by the discounted unit price. The FAQ states a minimum order quantity and unlimited users, and the add-ons are synthetic monitoring and logs (pay-as-you-go per GB). The two key points in reading it are that what is displayed is a "lower bound that includes the infrastructure-only plan", and that the full-stack unit price can only be pinned down through sales. For current unit prices, refer to Instana's official pricing page.
Datadog's Unit Price Is Determined by Three Tiers and Allotments
On the official pricing page, Datadog's APM shows the monthly amount per host when used together with Infrastructure Monitoring in three tiers - APM / APM Pro (bundles Data Streams Monitoring) / APM Enterprise (bundles Continuous Profiler) - each with annual, monthly, and on-demand unit prices side by side (Source: Datadog Pricing). The unit price for contracting APM alone does not appear on the public price list. What falls outside the unit price is the allotments: 150 GB of ingested spans per APM host per month and 1 million indexed spans per month (15-day retention) are included, and the allotments are evaluated as a pool across all hosts. Because overages are billed separately, the estimate needs "measured ingested GB per host" as the next input after "host count". How to choose among the three tiers is left to the explainer How to Think About APM Pricing, and how the SKUs stack up to How to Decode the Pricing Structure. This article explains how billing works. For current unit prices and plans, refer to each vendor's official pricing page.
The Variables That Decide Which Comes Out Ahead - What You Count and What You Multiply
When turning the difference in units into an actual estimate, the variables that move the monthly cost of the two can be organized into four. Which side comes out ahead flips depending on which variable is large in your environment.
| Variable | IBM Instana | Datadog | How it takes effect |
|---|---|---|---|
| Count (MVS / hosts) | Amount paid = MVS count × unit price. A minimum order quantity applies | Two lines: the Infrastructure Monitoring host price + the APM host price | Both scale with the count. For Datadog, compare using the combined "Infrastructure + APM" unit price |
| Tier chosen | Essentials (infrastructure only) or Standard (full stack). The unit price for Standard alone is not displayed | Three tiers: APM / APM Pro / APM Enterprise | Lining up only the displayed values compares Instana's lower bound against Datadog's combined value, which is not a like-for-like comparison |
| Span volume | No variable outside the unit price | Allotments for ingested GB and indexed count (pooled across all hosts); overages billed separately | Within the allotments, the count alone decides. If overage is the norm, only the Datadog side grows |
| Contract form | Annual contract / PayPerUse (hourly billing) / Self-Hosted | Annual contract / monthly contract / on-demand | Which form you use for the measurement period changes the premise of the comparison |
The reason Instana tends to look cheaper when only the surface lower-bound values are lined up is that Instana's "from" value is a lower bound that includes Essentials (infrastructure only), and the actual unit price when you choose the full-stack Standard is not shown on the page. Datadog's combined value includes the Infrastructure Monitoring host views, the container allotment, and the APM spans within the allotment. Only when you line them up again with the same count and the same content (full stack) does a comparison of unit price levels become meaningful. In a real estimate, obtaining the Standard unit price through sales for Instana, and measuring ingested GB per host in addition to host count for Datadog, are each indispensable steps.
Where Instana Wins
The first point is that the unit price has a single component. The footnote explicitly states that the amount paid is MVS count multiplied by unit price, with no second variable such as span ingestion volume or indexed count (as of September 2026, Source: Instana pricing page). The predictability that the bill will not jump for any reason other than host count even when traffic grows is something whose value is understood best by those who have operated Datadog's allotments. Next is instrumentation effort. The official documentation places automatic discovery and monitoring among the core capabilities and shows a design in which instrumentation comes through the agent, sensors, and AutoTrace (same date, Source: Instana SaaS documentation). In organizations where it is hard to get agreement to touch the application repositories, this difference decides whether adoption is feasible at all. The third is coexistence with the IBM product family: the same documentation has mainframe observability as a standalone chapter, and there is also the option of keeping data on-premises with Self-Hosted (an annual amount per MVS). Unlimited users and yen-denominated pricing are also practical benefits for procurement departments.
Where Datadog Pays Off: Tying What Lies Outside APM Together with the Same Tags
The reason to choose Datadog lies outside APM. Because logs, RUM, infrastructure metrics, and synthetic monitoring can be tied together with the same env / service / version tags as the APM traces, the round trip to what lies outside distributed tracing, "slow request, then the logs at that moment, then that host's CPU", is completed within a single screen. Without this round trip, incident investigation becomes the work of matching timestamps between tools. The three-tier pricing is also the freedom to buy only the depth you need. You can allocate APM Enterprise only to the services that need profiling and keep the rest on APM. Since instrumentation accepts OpenTelemetry in addition to the SDK, you can choose not to make your instrumentation vendor-specific. Instana is positioned for "APM quickly and broadly at a per-host price", and Datadog for "bundling all of observability with the same tags, starting from APM".
Three Things to Verify When Switching
Whichever direction you switch in, the three things to verify are the same.
- 1. Portability of instrumentation - Instana's FAQ states that, as a CNCF member, it can correlate open-source traces and metrics alongside its own collected data, and an Instana distribution of the OpenTelemetry Collector appears in the documentation. Datadog also has an OpenTelemetry intake, so if you instrument with OpenTelemetry, portability works in both directions
- 2. Converting the counting method - recount Instana's MVS (physical / virtual / worker node) and Datadog's host + container allotment + span allotment in the same ledger. The more container-centric the environment, the more the difference in counting methods shows up in the amount
- 3. Contract form - Instana is an annual contract with a minimum order quantity; Datadog has three forms: annual contract / monthly contract / on-demand. Decide first which vendor's on-demand you will use for the measurement period
Of the three, item 1 takes effect first. If your instrumentation depends deeply on a vendor-specific SDK, the effort of switching is determined not by the difference between tools but by the amount of code to rewrite. Conversely, if you are instrumented with OpenTelemetry, the comparison returns to the discussion of units in this article, and the only judgment left is which helps your own incident investigation more: Instana's simplicity or Datadog's breadth.