An Oracle Java SE to Azul migration is the cleanest exit when the customer needs a commercially supported OpenJDK distribution with predictable per-server pricing and zero Employee Metric exposure. Azul has been an OpenJDK commercial supplier for two decades and is the largest pure-play Java company outside Oracle. Azul Zulu is the free, TCK-certified OpenJDK build; Azul Platform Core is the paid commercial support contract; Azul Platform Prime is the differentiated pauseless GC product for low-latency workloads. This playbook lays out the buyer-side framework we use to displace the Java SE Universal Subscription onto Azul: the runtime decision per workload, the commercial negotiation, the forensic discovery, the deployment plan, and the audit defence on the historical Oracle Java footprint. Every step is benchmarked against real engagements — 600+ Oracle programmes, $1.8B in advised spend, and a 100% Java audit defence record.
Azul is the largest pure-play Java commercial vendor outside Oracle and has been shipping OpenJDK distributions since 2002. Azul Zulu has been TCK-certified for every OpenJDK release since Java 6, ships builds for every relevant CPU and operating system combination (Linux x64/aarch64, Windows x64, macOS x64/aarch64, Solaris, AIX, ARM32, MIPS, PPC), and is the OpenJDK distribution behind dozens of OEM embedded Java deployments. Azul's engineering organisation contributes back to OpenJDK across GC, JIT, security, and observability — and the company maintains a separate commercial codebase (Platform Prime) for differentiated low-latency workloads. This is not a re-badged binary; it is a first-party Java company with its own engineering investment.
The buyer-side reason to choose Azul over the bundled-support distributions (Microsoft Build of OpenJDK, Amazon Corretto) is commercial predictability for organisations without a qualifying Microsoft or AWS support relationship. Azul Platform Core pricing is a per-server, per-JVM, or per-developer commercial contract — not an Employee Metric. The contract is negotiable on volume, term, and metric. Azul also ships security CPUs four weeks ahead of OpenJDK GA, which is a material operational advantage for security-sensitive estates. The differentiated Platform Prime offering with the pauseless C4 garbage collector is the right answer for trading platforms, real-time analytics, and any workload where standard OpenJDK GC pauses are a measurable problem.
Azul is not always the right default. AWS-bundled support customers benefit more from Amazon Corretto. Azure-bundled customers benefit more from Microsoft Build of OpenJDK. Vendor-neutrality-required customers benefit more from Eclipse Temurin. The buyer-side methodology selects the runtime per workload. See the Oracle Java licensing guide for the Employee Metric exposure that drives the decision.
The Azul product portfolio has three distinct offerings and the per-workload decision determines the eventual commercial bill. Zulu is the free, TCK-certified OpenJDK build that any organisation can deploy at zero licence cost. Platform Core is the commercial support contract layered over Zulu — vendor-backed SLA, dedicated escalation engineering, ahead-of-GA security CPUs, and the supported binary stream. Platform Prime is the technically differentiated product with the C4 pauseless garbage collector, ReadyNow class pre-loading, and Falcon JIT compiler — targeted at workloads where standard OpenJDK GC is the operational bottleneck.
| Azul product | What it is | Where it fits |
|---|---|---|
| Azul Zulu | Free, TCK-certified OpenJDK distribution | Development workstations, non-critical workloads, community-support tolerance |
| Azul Platform Core | Commercial support contract over Zulu | Production workloads requiring vendor-backed SLA and ahead-of-GA CPUs |
| Azul Platform Prime | Commercial product with C4 GC, ReadyNow, Falcon JIT | Low-latency trading, real-time analytics, high-throughput services with GC pause problems |
The buyer-side methodology splits the estate into three tiers and applies a different commercial decision to each. Tier 1 (development workstations and non-critical workloads) gets free Azul Zulu with no commercial contract. Tier 2 (production workloads requiring vendor support) gets Azul Platform Core. Tier 3 (a small population of latency-sensitive workloads where the business case justifies the premium) gets Azul Platform Prime. Most enterprise estates land at 60 to 70 percent on free Zulu, 25 to 35 percent on Platform Core, and 0 to 10 percent on Platform Prime. The structural commercial outcome is a multi-tier estate with predictable per-server economics, against the Oracle Employee Metric which charges full price for every employee regardless of actual deployment.
Buyer-side note: Azul's sales team will push for a flat per-employee or per-VM contract that simplifies their commercial picture. Push back. The per-server (or per-JVM) metric is the one that aligns Azul's cost to your actual deployment, and the one that protects you from headcount growth. Negotiate the metric explicitly before the price.
We benchmark your Azul proposal against the alternative OpenJDK distributions and against published Platform Core / Platform Prime price points. Buyer-side methodology, ten-day turnaround.
Azul is a commercial vendor with a sales motion, and the contract terms matter. The negotiation framework we apply against Azul is the same buyer-side framework we apply against Oracle, with a different fact base. The framework defends against the four commercial risks of any Azul commercial contract: metric drift, headcount-based pricing, term escalation, and exit obstruction. Each risk is contestable in negotiation if the customer enters the discussion with a documented baseline and an alternative.
The benchmark price points we see across engagements: Azul Platform Core typically lands at $200 to $600 per server per year for standard enterprise SLA coverage; Azul Platform Prime typically lands at $1,500 to $3,500 per server per year for the differentiated pauseless GC. The total commercial spend on Azul for a typical mid-size enterprise estate (500 to 2,000 servers) lands at $300K to $1.2M per year — against an Oracle Java SE Employee Metric exposure of $2M to $8M per year for the same population. The structural saving justifies the migration; the negotiation framework right-sizes the eventual Azul bill. The Contract Negotiation service covers the framework end to end.
The forensic discovery phase is identical across all OpenJDK migration targets. The audit defence framework requires a complete, evidence-based inventory of every Oracle Java installation in the estate, identified by vendor, version, deployment context, and licensing surface. Most organisations underestimate their Oracle Java footprint by a factor of three to ten because the binary is bundled inside dozens of third-party products and gets installed by default by package managers and developer tooling.
| Discovery surface | What to look for | Audit exposure |
|---|---|---|
| Endpoint inventory (workstations) | Oracle JRE / JDK on Windows / macOS / Linux desktops | Employee Metric applies regardless of usage |
| Server-side application servers | Tomcat, Jetty, JBoss, WAS, custom JVMs | Production Java SE installations |
| Low-latency trading platforms | Java-based trading engines, market data, risk | Frequent Platform Prime candidate; high audit value |
| Batch and ETL processing | Informatica, DataStage, custom Java batch | Server-side Java SE installations |
| Build and CI tooling | Jenkins, GitLab Runner, Bamboo, Maven, Gradle | Developer tooling Java; Oracle binaries trigger metric |
| Third-party software with bundled Java | Trading platforms, BI tooling, scientific applications | Oracle Java bundled inside vendor packages |
| Container images | Docker / OCI images on private registries | Each image with Oracle Java multiplies the surface |
| OT and ICS environments | SCADA, manufacturing, embedded engineering tooling | Frequently overlooked; high audit value |
The forensic discovery has to be performed under audit-defensible methodology. We combine endpoint scanning, package manager inventories, container registry scans, third-party vendor inventories, and structured interviews of OT / engineering teams to surface the embedded footprint. The output is the Java installation register that drives both the migration plan and the audit defence. The discovery also identifies the Platform Prime candidate workloads — low-latency trading, real-time analytics, large-heap services with measured GC pause problems. The Java Licensing service covers the discovery framework end to end.
The Azul deployment follows the same wave-based sequence we use for every OpenJDK migration, with one additional phase: the Platform Prime evaluation phase for the small population of latency-sensitive workloads. The standard wave-based deployment puts free Azul Zulu on Tier 1, Azul Platform Core on Tier 2, and the Platform Prime evaluation on the candidate Tier 3 workloads. The Platform Prime evaluation is a structured benchmark against the existing Oracle binary and standard OpenJDK on the same workload — the only honest way to decide whether the premium is justified.
| Phase | Duration | Activity |
|---|---|---|
| 1. Pilot deployment | 2 weeks | Deploy Azul Zulu on 3 to 5 non-critical workloads; functional validation |
| 2. Performance baseline | 2 weeks | Compare GC behaviour, throughput, latency against Oracle baseline |
| 3. Platform Prime evaluation | 4 weeks | Benchmark Platform Prime on candidate latency-sensitive workloads |
| 4. Wave 1 deployment | 4 weeks | Deploy Zulu / Platform Core to 20 to 30 percent of low-risk workloads |
| 5. Wave 2 deployment | 6 weeks | Deploy to mid-risk workloads; retain Oracle binary fallback |
| 6. Wave 3 deployment | 8 weeks | Deploy to high-risk workloads; full rollback plan; Platform Prime production |
| 7. Endpoint rollout | 4 weeks | Workstation replacement via endpoint management tooling |
| 8. Oracle Java removal | 3 weeks | Forensic removal of Oracle binaries; verification scan |
| 9. Audit-defensible documentation | 2 weeks | Evidence package for the future Oracle audit defence |
The Platform Prime evaluation phase is the one we recommend customers run with the most rigour. The Platform Prime commercial premium is justified only when the workload demonstrably benefits from the pauseless GC — typical measurable benefits include sub-10ms 99.99-percentile latency, elimination of GC-related throughput dips, and the ability to run substantially larger heaps without pause problems. Where the benefit is not measurable, Azul Platform Core on the standard Zulu binary is the correct commercial decision and the Platform Prime line item should be removed from the contract.
We sequence the Azul wave-based deployment, run the Platform Prime evaluation, and right-size the eventual Azul commercial contract. Buyer-side methodology, ten-day turnaround.
The migration to Azul closes future Oracle Java SE exposure. It does not close the historical exposure. Oracle's GLAS / LMS Java audit team reviews the customer's historical Oracle Java SE deployment to determine retroactive Employee Metric exposure and publishes a back-licence claim. The defence framework is the same regardless of which OpenJDK distribution the customer migrated to — the forensic defence sits alongside the migration, not after it.
The forensic defence framework contains four lines of argument. First, the licensing model in effect at the time of deployment was the historical metric (Processor or NUP), not the 2023 Employee Metric — Oracle's retroactive application is contestable. Second, the actual Oracle binary deployment count, not the inferred deployment from desktop installation telemetry, is the basis for any back-licence claim — Oracle's audit methodology routinely overstates by a factor of two to four. Third, the Oracle Java No-Fee Terms and Conditions (NFTC) that applied to JDK 17 between September 2021 and September 2024 created a covered-deployment window that excludes those deployments from back-licence claims. Fourth, the bundled-Java distributions of Oracle Java that came with Oracle products (Database, WebLogic, Fusion Middleware) carry application-specific entitlement and cannot be treated as standalone Java deployments.
We hold a 100% Java audit defence record across the Java engagements we have run since 2019. Anonymised case: a North American investment bank moved its trading-platform Java estate from Oracle Java SE to Azul Platform Prime and the rest of the estate to Platform Core; the Oracle audit follow-up was challenged on the historical metric and the inferred deployment count, with the final settlement reduced by 41% from the opening Oracle position and converted to a one-off non-renewing exit fee. The Audit Defence service covers the full engagement framework. The Oracle audit guide covers the broader evidence-based methodology.
Azul Zulu is the free, TCK-certified OpenJDK build. Azul Platform Core is the paid commercial support contract for Zulu, with vendor-backed SLA coverage, security CPUs delivered four weeks ahead of OpenJDK GA, and dedicated engineering escalation. Azul Platform Prime is the differentiated commercial product — a pauseless garbage collector (C4) targeted at low-latency, high-throughput workloads where standard OpenJDK GC pauses are a measurable problem. Most enterprise estates need Zulu plus Platform Core; Platform Prime is a workload-by-workload commercial decision driven by latency budgets. The Oracle Java SE to Azul migration typically splits the estate across all three tiers.
Azul Platform Core is typically priced per server, per JVM, or per developer rather than on the Oracle Employee Metric. The pricing model is the major commercial advantage: Azul charges for what you run, not for every employee in the organisation. Real-world Platform Core pricing typically lands between 20 and 40 percent of equivalent Oracle Java SE Employee Metric pricing for similar enterprise SLA coverage. The Azul commercial contract is also negotiable on volume, term, and metric — unlike the Oracle Employee Metric which Oracle treats as non-negotiable.
Azul is the right default in three scenarios. First, when commercial vendor-backed support is required but the customer has no qualifying Microsoft or AWS support relationship. Second, when the customer wants the security CPUs delivered ahead of the OpenJDK GA release schedule (Azul ships them four weeks earlier). Third, when specific workloads need Platform Prime's pauseless GC for low-latency or high-throughput requirements. Microsoft Build of OpenJDK is the right default for Azure-bundled estates; Amazon Corretto for AWS-bundled estates; Eclipse Temurin for vendor-neutral requirements.
Azul is a private company headquartered in California with two decades of OpenJDK contribution history. The single-vendor risk is real but bounded — if Azul commercial support became unavailable, the underlying Zulu binaries are TCK-certified OpenJDK and the customer can migrate to Eclipse Temurin, Amazon Corretto, or Microsoft Build of OpenJDK on the same Java SE bytecode contract. The exit risk is operational, not technical. We recommend that any Azul commercial contract include explicit source-availability clauses and an exit-assistance clause to preserve the migration option.
The audit defence framework is the same regardless of which OpenJDK distribution the customer migrated to. Oracle's GLAS team examines the historical Oracle Java SE footprint, applies the Employee Metric retroactively, and publishes a back-licence claim. The defence framework reduces or eliminates the claim through four lines of argument: contesting the retroactive application of the new metric, contesting the inflated installation count from telemetry, applying the NFTC covered-deployment window for JDK 17, and excluding bundled-Java distributions that carry application-specific entitlement. We hold a 100% Java audit defence record across engagements run since 2019.
Twice a month. Oracle pricing moves, audit-defence tactics, migration economics, ULA exit playbooks. Written by former Oracle insiders, never marketing.
No spam. Unsubscribe any time. Independent — not affiliated with Oracle Corporation.