Runtime support

Java Support, 24/7, with Emergency Response for Production JVMs

Last reviewed by Tyler Eastridge

AceMQ supports Java 24/7 across Americas, EMEA and APAC for production JVMs on Oracle JDK, OpenJDK, Eclipse Temurin, Amazon Corretto and Azul Zulu, on virtual machines, bare metal and Kubernetes. Most calls start with one of three problems: a pod killed with a healthy heap, garbage collection pauses that break latency targets, or TLS failures after a JDK change.

24/7 Java support with response targets from 15 minutes, by plan. Quote replies within 1 hour.

Support quotes answered within 1 hour, by a senior engineer.

Planning a build, upgrade or migration instead? See AceMQ Java consulting and services.

160+
Customers supported
10+
Years in production support
40+
Sev 1 incidents handled a year
24/7
Americas, EMEA and APAC
1hour
Quote reply

Customers

  • Microsoft
  • Henry Schein
  • Enbridge
  • Gojek

Partners

  • Broadcom
  • Redis
  • VMware Tanzu
  • OSSeva

Incident Triage

Problems we fix every week

Most Java incidents come down to what the JVM believes about its container. These rows start from jcmd output, GC logs or thread dumps.

Common Java production problems and how AceMQ resolves them
Symptom you're seeingWhat we check first
High CPU in a container that is mostly idleWe compare the container CPU quota with what the JVM detects (java -XshowSettings:system -version), then -XX:ActiveProcessorCount and the GC and compiler thread counts it implies. Then a two-minute profile: jcmd <pid> JFR.start duration=120s settings=profile filename=/tmp/app.jfr, read for hot methods, lock contention and allocation.
unable to create native threadThe limit is not memory. We check ulimit -u, the container PID limit, grep Threads /proc/<pid>/status, and jcmd <pid> Thread.print for pools that grow without bound. A thread-per-request design needs a bounded executor or virtual threads, and the fix goes in the code path we identify.
Deadlocks and pinned virtual threadsjcmd <pid> Thread.print reports "Found one Java-level deadlock" with the owner chain. On virtual threads we run with -Djdk.tracePinnedThreads=full or read the JFR jdk.VirtualThreadPinned events to see which synchronized block holds a carrier.
Direct memory and NettyOutOfMemoryError: Direct buffer memory points to -XX:MaxDirectMemorySize or a leak. We set -Dio.netty.leakDetection.level=advanced in a test environment, read Native Memory Tracking (-XX:NativeMemoryTracking=summary with jcmd <pid> VM.native_memory summary), and size the container from heap plus direct plus metaspace plus thread stacks, not heap alone.
Pauses after a collector or JDK changeWe read -Xlog:gc*,safepoint:file=gc.log against request latency. G1 settings, generational ZGC (-XX:+UseZGC, with -XX:+ZGenerational on JDK 21) and Parallel for small containers have different costs, and the decision follows allocation rate, heap size and the latency target.
Leak hunting in productionjcmd <pid> GC.class_histogram twice, ten minutes apart, shows what grows. Then jcmd <pid> GC.heap_dump and a dominator-tree read, or the JFR old object sample event when a full dump is too disruptive.
TLS fails after an updateBeyond the truststore, we check jdk.tls.disabledAlgorithms in java.security for protocols and key sizes that a newer update disabled, and run with -Djavax.net.debug=ssl:handshake to see which side rejected which parameter. keytool -list -cacerts shows what the process really trusts.
An upgrade is planned and nobody knows what will breakjdeps --print-module-deps and jdeprscan list module and removed-API use; the JDK 17 strong encapsulation check is run with --add-opens flags logged, not guessed. We stage Java 8, 11, 17, 21 and 25 moves in that order of risk, with tests at each step. See /tomcat/support/ and /spring-boot/extended-support/ for the layers above the JVM.

Each production incident ends with an RCA naming the flag, limit or code path behind it and the setting we changed.

JVM keeps getting killed or pausing for seconds? Request a quote and we reply within the hour.

Contact Us

Deployment coverage

Which Java deployments does AceMQ support, and who covers what?

DeploymentWhat AceMQ coversWhat the vendor coversExcluded
Oracle JDK with a Java SE SubscriptionProduction troubleshooting, tuning, upgrade planning, written root cause analysisOracle supports its JDK and supplies updates to subscribers (Oracle Java SE Support Roadmap, updated 20 Oct 2026, read 17 Nov 2026)JDK defects stay with Oracle
OpenJDK builds: Eclipse Temurin, Amazon Corretto, Azul Zulu, Red Hat and Microsoft buildsThe same scope, on whichever build you runThe build's vendor supplies updates; some sell supportDefects in a vendor's build stay with that vendor
Java in containers and KubernetesMemory and CPU limits against JVM detection, probes, GC, startup, native memoryThe platform vendor supports KubernetesControl-plane faults
Java on application servers and frameworksThe JVM layer, and escalation to the matching page for the layer aboveSee /tomcat/support/ and /spring-boot/extended-support/Application code rewrites

Support options

Who provides enterprise support for Java? Oracle, build vendors and independent support compared

The vendor of your JDK build supplies the runtime and its security updates, and some vendors sell support for it. Independent support supplies the engineers who run and troubleshoot the JVM in production, whichever build it is.

Oracle / Amazon Corretto / Eclipse Temurin
AceMQ independent support
What you getOracle: Oracle Java SE Subscription and support for Oracle Customers. Java 8, 11, 17, 21 and 25 are LTS releases; Oracle JDK 25 is free for all users under the No-Fee Terms; updates for 17 and 21 are under the OTN licence, free for personal and development use only (Oracle roadmap, read 17 Nov 2026)Amazon Corretto: Free LTS builds; updates quarterly; no Corretto-specific assistance plans, though an AWS Support plan can include Corretto help (Corretto FAQ, read 17 Nov 2026)Eclipse Temurin: Free builds; Eclipse states it does not provide paid support and points to a commercial support list (Adoptium support page, read 17 Nov 2026)
What you get24/7 production engineers for any build
Response commitmentOracle: Per the Oracle support contractAmazon Corretto: Per the AWS support planEclipse Temurin: None
Response commitmentResponse targets from 15 minutes, by plan
ScopeOracle: Oracle productsAmazon Corretto: Corretto buildsEclipse Temurin: Temurin builds
ScopeYour JVM, tuning, incidents, upgrades
PriceOracle: By quoteAmazon Corretto: FreeEclipse Temurin: Free
Price$24,000 to $72,000 a year (USD), set by SLA, scope and deployment

What's Included

Everything in your Java support contract

Runtime behavior on any distribution: memory and GC, threads, TLS, upgrades between versions, and the apps running on top of them.

  • 01

    Emergency Incident Response

    A JVM that is down, crash-looping, out of memory or failing every request. With response targets from 15 minutes, by plan, an AceMQ engineer joins a live bridge, 24/7, to diagnose from heap dumps, GC logs, thread dumps and crash files, not to acknowledge a ticket.

  • 02

    Root Cause Analysis

    Every P1 closes with a written RCA: what failed, why, the fix applied and the setting or code change that prevents a repeat. Delivered as standard, not on request.

  • 03

    LTS Upgrade Planning

    Moves from Java 8 or 11 to 17, 21 or 25, with the library, framework and flag changes found and rehearsed on a copy before production. Frameworks set floors: Spring Boot 3 needs Java 17, and Apache Kafka 4.0 brokers need Java 17.

  • 04

    Lifecycle and Exposure Advisory

    Which of your JDK builds are inside their vendor's support window, which are not, and what that means for security updates. Dates are read from each vendor's own page, not from a summary.

  • 05

    Memory, GC and Container Review

    Heap, GC and thread settings checked against the container's real memory and CPU limits, with Native Memory Tracking to find what lives outside the heap.

  • 06

    TLS, Certificate and Security Review

    Truststores, default security properties and disabled algorithms checked across the JDK builds in use, so the next base image change does not break a handshake.

Why AceMQ

What our Java engineers do differently

  • Engineers Who Know Your Environment

    Tickets go to senior engineers briefed on which services run which JDK, so a P1 starts with a hypothesis rather than an interview. The proposal states who does the work.

  • No Help-Desk Layer

    You reach a senior engineer directly by phone, email or Slack. No queue collecting details to relay, and no approval between you and someone who can read a heap dump.

  • Neutral on the Distribution

    AceMQ sells no JDK, so it has no reason to steer you to Temurin, Corretto, Zulu, the Red Hat build, the Microsoft build or Oracle. It supports the one you run.

  • Honest About Patches

    AceMQ produces no JDK patches and says so before you sign. A defect that needs a JDK code fix goes to the build vendor or the OpenJDK project, and we help you build the case.

  • The Layers Above the JVM

    Most Java incidents involve the application server or framework as much as the JVM. The same team covers Tomcat and Spring Boot, so the diagnosis does not stop at the runtime boundary.

  • One Contract Across Your Stack

    Java sits next to Kafka, RabbitMQ, Elasticsearch and PostgreSQL in most estates. One AceMQ contract and SLA can cover them, with multi-technology discounts.

JVM platforms we support

Oracle JDK and OpenJDK builds from Temurin, Corretto, Zulu, Red Hat and Microsoft, on VMs, bare metal and Kubernetes.

  • Java 8, 11, 17, 21 and 25 LTS
  • Oracle JDK and OpenJDK builds
  • Eclipse Temurin, Amazon Corretto and Azul Zulu
  • Red Hat build and Microsoft Build of OpenJDK
  • Virtual machines
  • Bare-metal servers
  • Docker containers
  • Kubernetes and OpenShift
  • AWS, Azure and Google Cloud instances you run
  • Applications on Tomcat, Spring Boot and Kafka brokers

Version coverage

Versions we support

Versions we support
Java lineEclipse Temurin: planned updates until at leastAmazon Corretto: last planned update
Java 25 (LTS)Sep 2031Jul 2032
Java 21 (LTS)Dec 2029Jul 2030
Java 17 (LTS)Oct 2027Jul 2029
Java 11 (LTS)Oct 2027Oct 2031
Java 8 (LTS)Dec 2030Oct 2030

Sources: Adoptium support page and Amazon Corretto FAQ, both read 17 November 2026. Dates are each vendor's own and differ by build, so check yours before you plan. AceMQ supports all five lines in production and plans moves between them.

Where support stops

What Java support covers and what stays with your JDK vendor

Support covers the people and process around a running JVM. It does not cover the JDK's own patch stream, and a page that blurs the two sends buyers into the wrong decision.

  • What AceMQ provides

    Engineers who diagnose, tune and upgrade the runtime in production, on-call response, root-cause analysis, upgrade planning and lifecycle advice, under one AceMQ contract and SLA.

  • What stays with your JDK vendor

    The JDK build and its security updates. Oracle, the Eclipse Adoptium project, Amazon, Red Hat, Microsoft and Azul each publish their own, on their own dates. AceMQ produces none.

  • If a defect is in the JDK

    A crash or wrong result that traces to the JDK itself needs a code fix from the build vendor or the OpenJDK project. We reproduce it, write it up and raise it with the right party.

  • If a release is past its window

    When your vendor stops updating a release, the choices are to move to a newer LTS, to buy that vendor's extended support, or to treat the host as an exception with controls around it. We lay out which fits your estate.

Dates and licensing terms change, so this page reads vendor dates from each vendor's own page. Licensing questions belong to your agreement and your counsel; AceMQ gives no licensing advice. End-of-life software support

Escalation Path

Your first hour of a Java outage

Whoever picks up a JVM that keeps getting killed has read GC logs and native memory reports before, so the first hour goes to evidence.

  1. From 15 min (by plan)first responseEmergencyProduction service down or restarting, container killed with heap not full, JVM deadlocked under load
  2. 1 hourfirst responseCriticalNative thread exhaustion, direct memory errors, long GC pauses hitting request latency
  3. 4 hoursfirst responseHighTLS failures after a JDK update, CPU burn in an idle container, suspected leak in production
  4. Next Dayfirst responseStandardQuestions on collector choice, JFR use, distribution selection and version upgrade planning

Response follows your plan. A heap dump of a leaking service in production and a GC tuning question get different clocks.

Want a named Java engineer across your services? Get a quote or write to us.

Contact Us

Pricing

What does AceMQ Java support cost?

Java support runs about $24,000 to $72,000 a year, quoted per scope. It depends on how many applications and JVM hosts matter in production, how many distributions you mix, and the response level.

Prices are in USD. Response targets start at 15 minutes depending on plan, 24/7, and quote requests are answered within 1 hour. Adding Tomcat, Spring Boot or Kafka support to the same contract earns a discount.

Typical engagements

Java incidents we have untangled

Java engagements where container limits, thread exhaustion or a version move was diagnosed and settled.

1 / 6 case studies

  • Assessment

    Java Estate Inventory and OpenJDK Distribution Selection

    Mid-Sized Distribution Company

    A typical engagement for a company that wants to leave an Oracle Java SE subscription or avoid one: inventory every Java runtime, compare OpenJDK distributions against the estate, and write the migration plan.

    JavaOpenJDKDockerKubernetes
  • Assessment

    Datadog APM Instrumentation Coverage Assessment

    Global Payments Processor

    Assessment of APM instrumentation coverage and trace completeness across a service estate where distributed traces kept breaking at service boundaries.

    DatadogOpenTelemetryKubernetesJavaPython
  • Remediation

    New Relic Agent Overhead Remediation

    Digital Commerce Platform

    Remediation of application latency and memory growth traced to APM agent configuration on the customer's own JVM and container estate.

    New RelicJavaKubernetesLinux
  • Support

    New Relic Distributed Tracing Support

    Insurance Services Provider

    Ongoing support for distributed tracing across a mixed New Relic and OpenTelemetry estate where traces broke at instrumentation boundaries.

    New RelicOpenTelemetryJava.NETKubernetes
All use cases

Support options

Four tiers of Java support.

AceMQ offers Java support in four tiers, each handing more of the day-to-day work to AceMQ: 24/7 support and emergency response for JVMs your team runs, proactive reviews that find memory, TLS and lifecycle problems first, managed Java runtimes on your own infrastructure, and managed Java runtimes on AceMQ infrastructure. Java 8, 11, 17, 21 and 25 are covered by AceMQ engineers working with partner Java specialists, on Oracle JDK or any OpenJDK build. JDK patches still come from your distribution vendor.

  • Tier 1

    Java 24/7 Support & Incident Management

    Your developers own the JVMs, we step in

    Best for: Application teams who handle day-to-day Java and need a specialist for OOM kills, pauses and thread issues.

  • Tier 2

    Java 24/7 Support with Proactive Monitoring

    We tune flags, GC and upgrade readiness

    Best for: Estates running mixed JDK builds where nobody has checked container awareness or deprecated API use.

  • Tier 3

    Managed Java on Your Infrastructure

    We operate JVM platforms you host

    Best for: Organizations that need runtimes, patches and monitoring kept on their own servers and clusters.

  • Tier 4

    Fully Managed Java on AceMQ Infrastructure

    We run your JVM application tier for you

    Best for: Teams who want applications running without staffing JVM patching and diagnostics.

FAQ

Java support questions

Still have questions about Java? Email an Expert

Who provides enterprise support for Java if we do not buy Oracle Java SE?

AceMQ does, whichever JDK distribution you run, at any hour. On the top plans a response begins within 15 minutes. A Sev 1 could be a container killed repeatedly while the heap looks half empty: we read Native Memory Tracking and the cgroup limit, then deliver an RCA. P2 1 hour, P3 4 hours.

Why is our Java container killed when the heap is not full?

The container limit includes metaspace, thread stacks, code cache and direct memory. Native Memory Tracking shows the real split, and the limit and -XX:MaxRAMPercentage follow from that.

Which garbage collector should we use?

It follows heap size, allocation rate and your latency target. We read your GC logs and test the candidate against production-like load before changing the default.

How do we find a memory leak in production?

Two class histograms ten minutes apart show what grows, then a heap dump or JFR old object sample shows what holds it.

Can you support Java 8 and 11?

Yes, as production runtimes, and we plan moves to 17, 21 or 25 from the version and build you run.

Why did TLS break after a JDK update?

A newer update disabled a protocol or key size, or the new JDK has its own truststore. Run with handshake debugging and compare cacerts.

Do you also support Tomcat and Spring Boot?

Yes, on their own pages. This page covers the JVM layer; Tomcat and Spring sit above it, and one contract can include all three.

What does a Java health check include?

Heap, GC and native memory review, container limits against JVM detection, thread and pool configuration, TLS and truststore inventory, build and update plan, with a ranked list of fixes.

Can we buy Java help for one project only?

Yes. A single engagement, for instance a memory leak hunt, a GC tuning study or an upgrade readiness check using jdeps, needs no yearly term. Many teams then take an annual plan for 24/7 response and the same engineers across their JVM estate.

How is Java support priced?

Contracts commonly fall between $24,000 and $72,000 a year (USD). The factors are the number of production applications and JVM hosts, the mix of JDK distributions and versions, whether Tomcat, Spring Boot or Kafka are in scope, and how fast you need a reply. Quotes return within 1 hour.

Let's talk

Know Who Answers When the JVM Goes Down

Send us a GC log and the container spec of the service that worries you. We reply within 1 business hour with what we see.

A senior Java engineer, not a sales queueQuote requests answered within 1 hourAd hoc engagements or an annual partnership

Production incident right now?Email a Java Architect