HHVM

HHVM
Name	HHVM
Developer	Meta Platforms
Released	2010
Written in	C++, PHP, Hack
Operating system	Linux, macOS
License	PHP License, MIT

HHVM is a virtual machine and just-in-time (JIT) compiler project originally developed by engineers at Meta Platforms. It was created to improve execution speed for server-side scripting languages and to evolve runtime techniques used in large-scale web infrastructures. HHVM influenced language design, runtime optimization, and deployment strategies across prominent technology organizations and research institutions.

History

HHVM emerged from work by engineers at Facebook, who drew on lessons from projects at Google, Microsoft Research, Yahoo!, Amazon Web Services, and IBM Research to address performance issues in large PHP codebases. Early development coincided with initiatives at Sun Microsystems and the Linux Foundation exploring runtime performance; contributors cited prior art such as V8 (JavaScript engine), HotSpot, and OpenJDK. As adoption grew, HHVM intersected with initiatives at Facebook AI Research and collaborations with teams from Instagram, WhatsApp, and Oculus VR. The project attracted attention from academic groups at MIT, Stanford University, UC Berkeley, and Carnegie Mellon University studying dynamic language optimization. Over time, leadership and contributions involved engineers with ties to LinkedIn, Dropbox, Pinterest, and Airbnb. Licensing discussions referenced models used by Mozilla Foundation and Apache Software Foundation projects. HHVM’s trajectory paralleled trends at Oracle Corporation and Red Hat, while developers engaged with package ecosystems overseen by Debian, Ubuntu, and Homebrew maintainers.

Architecture and Components

The architecture integrated a high-performance JIT compiler informed by research from Bell Labs and publications from ACM and IEEE. Core components include a bytecode emitter influenced by Zend Engine implementations and a trace optimizer drawing on techniques from HP Labs and Oracle Labs. The runtime uses memory management strategies comparable to those in V8 (JavaScript engine), with garbage collection designs examined alongside work from Benoit Mandreuil-era studies and Paul Graham-style Lisp systems. HHVM’s componentization mirrored service designs used at Twitter and Netflix, with subsystem interfaces patterned after gRPC and Thrift architectures originating at Google and Facebook. Language frontend integration involved teams familiar with Zend Engine, Phabricator, and tools from GNU Project. Build systems and continuous integration pipelines leveraged tooling from Jenkins, Travis CI, and CircleCI while packaging referenced conventions from RPM, apt, and Homebrew.

Performance and Benchmarking

Benchmarks comparing HHVM referenced suites similar to those published by SPEC, Phoronix Test Suite, and research groups at ETH Zurich and Princeton University. Early claims of throughput improvements were evaluated against optimizations in V8 (JavaScript engine), PyPy, and LuaJIT, and measured on hardware from Intel Corporation and AMD. Performance discussions involved database-backed workloads using MySQL, PostgreSQL, and Memcached, with caching layers like Redis and CDNs such as Akamai. Webserver integrations for benchmarks included nginx, Apache HTTP Server, and Lighttpd deployments. Real-world telemetry from deployments at Facebook, Flickr, and WordPress.com informed tuning guidance, while academic comparisons cited papers from SIGPLAN, PLDI, and OOPSLA conferences.

Compatibility and Language Support

HHVM supported a dialect introduced by Meta engineers to extend features familiar to developers who used PHP, and interaction with ecosystems like Composer and Packagist shaped compatibility efforts. Integration with frameworks and platforms including Laravel, Symfony, Drupal, Magento, MediaWiki, and WordPress required adaptation. Tooling compatibility involved editors and IDEs such as Visual Studio Code, PhpStorm, Sublime Text, and Vim. Interoperability with data serialization standards from JSON, XML, and Protocol Buffers and RPC systems like Thrift influenced language bindings. Language design discussions referenced paradigms popularized by Martin Fowler, Kent Beck, and Bjarne Stroustrup in related contexts.

Deployment and Use Cases

HHVM was deployed in large-scale web services and backend platforms operated by teams with infrastructure influenced by practices from Google SRE, Netflix OSS, and Pinterest Engineering. Typical stacks combined HHVM with reverse proxies like Varnish, service meshes inspired by Istio, and orchestration systems such as Kubernetes and Docker Swarm. Use cases included social networking backends similar to Facebook, content management systems like WordPress.com, SaaS offerings from companies akin to Shopify, and high-throughput APIs used by Stripe-like payment services. Integration patterns matched logging and monitoring solutions from Splunk, Datadog, Prometheus, and ELK Stack components developed by Elastic NV.

Security and Development Community

Security practices around HHVM referenced standards and advisories from OWASP, CVE Program, and guidance used by CERT Coordination Center. Vulnerability disclosure and patching workflows paralleled processes used by Mozilla Foundation and Apache Software Foundation projects. The development community included contributors with experience at Meta Platforms, Google, Microsoft, Red Hat, and independent maintainers from open source collectives associated with GitHub and GitLab. Governance discussions echoed models from Linux Foundation projects and foundations like Apache Software Foundation and Mozilla Foundation. Academic collaborations and conference presentations occurred at venues such as USENIX, ACM SIGOPS, EuroSys, and FOSDEM.

Category:Virtual machines