PayPal’s clearest explanation of how Linux and open-source software paid off comes from a 2007 account by then-vice president of core technologies Matthew Mengerink. He described practical benefits—not a published company-wide savings figure: fewer licensing barriers to replicating production in a test lab, more flexible use of server capacity, and security controls tailored to PayPal’s environment. Later PayPal accounts of OpenStack, HERA, and Kafka show related infrastructure choices, but they do not establish that PayPal uses the same architecture today.
What PayPal said Linux made possible
In 2007, Mengerink described an in-house environment of “thousands of Linux-based, single-rack Unity servers.” Those machines supported the web-presentation layer, middleware, and user interface. This is a historical snapshot, not a verified description of PayPal’s current production estate. LinuxInsider’s 2007 account is the most direct source for the original business case.
The payoff he described came from how the company could use and operate the infrastructure. Linux and open-source software gave PayPal a platform for replicating environments, moving workloads, and applying its own operational policies. The account does not compare total ownership costs against a proprietary alternative, or quantify net savings.
Lower licensing exposure and more consistent testing
Mengerink said PayPal could reproduce its live site in a lab without incurring high licensing charges. Matching development and production environments also made tests more likely to behave as expected and helped developers diagnose problems. In his words, “When developers are testing things in the same environment that we actually have in production, we’re far more likely to get a consistent and expected result.”
The Tool Desk
Outbyte Driver Updater FREEScan for outdated or missing drivers - takes under a minuteDriver Scan →Outbyte PC Repair FREERepair Windows errors before they cause bigger problemsFix Now →#1 Best Overall
That is a plausible cost and development mechanism: avoiding some license charges for replicated environments while reducing differences between test and production. It is PayPal’s stated rationale, not an independently measured savings result. The account does not provide a complete cost model that includes staffing, support, security work, or ongoing operations.
Flexible servers and inexpensive redundancy
PayPal’s 2007 description also emphasized workload flexibility. Middle-tier Linux servers could be reassigned to run daily payment-reconciliation batches, which Mengerink said could finish “in a couple of hours.” The example shows how spare or repurposed capacity could serve more than one function rather than remain dedicated to a single workload.
Rank #2
Mengerink also argued that inexpensive redundancy made a distributed system practical, contrasting it with reliance on a large, less flexible system. That was his explanation of PayPal’s design choice, not proof that distributed systems are always cheaper or more reliable. Distributing workloads also creates operational responsibilities: teams must manage failures, coordination, capacity, and the complexity of many machines.
Security required deliberate operations
PayPal’s account did not treat open-source software as secure by default. Mengerink described tailored security policies and said the company treated machines as though they were on an untrusted network. He also mentioned Red Hat kernels with custom changes.
Recommended Free Tools
The lesson is that the platform’s flexibility came with responsibility for configuration and security practices. Access control, patching, network assumptions, and integration remain operational work; source availability alone does not guarantee a secure system.
Later examples: OpenStack, HERA, and Kafka
Subsequent PayPal accounts describe other infrastructure decisions that illustrate related themes—shared projects, internal development, and operating open-source platforms at scale. They provide context, not evidence that the exact 2007 environment continued unchanged.
| Example | What PayPal reported | What the evidence does not establish |
|---|---|---|
| OpenStack, 2014 | The Open Infrastructure Foundation case study described a private-cloud effort intended to improve agility, availability, and innovation. PayPal stated a requirement of 99.9999% availability. | The figure is a design requirement, not proof of achieved uptime or a description of PayPal’s current cloud deployment. |
| HERA, 2019 | PayPal said it built a database-access gateway to address connection-pooling and scaling needs after not finding a suitable commercial or open-source option for its requirements. It later released HERA under Apache 2. The post reported that an early version reduced connection latency from triple-digit to single-digit milliseconds. | The latency result is PayPal’s report about one system; it is not an overall financial-return measure. |
| Kafka, 2023 | PayPal’s engineering blog described a Kafka platform supporting mission-critical applications and ingesting “trillions of messages per day.” It discussed configuration, access control, monitoring, and automation, as well as PayPal’s contribution of KIP-519 for extensible SSL context and engine behavior in certificate handling. | The scale figure is company-reported, not independently audited, and does not prove the same scale or architecture applies to every PayPal system. |
In 2014, senior director of PayPal infrastructure engineering Saran Mandair said, “By leveraging the collective innovation of the OpenStack community, we can develop and grow our private cloud much quicker without having to reinvent anything.” The point is a potential benefit of building on shared infrastructure: a company can use community-developed capabilities and contribute findings back rather than develop every component alone. PayPal’s Kafka account offers a later example of upstream participation through KIP-519. Neither example shows that all PayPal-specific code was released publicly.
HERA illustrates a different route: when PayPal said available products did not meet a particular database-access requirement, it developed a system internally and then released it as an Apache 2 project. Its reported early latency improvement is an operational result for that system, not evidence of company-wide savings.
What these examples can—and cannot—say about return on investment
The case for payoff is strongest as a set of mechanisms and operational examples: fewer licensing barriers to copying environments, more consistent testing, the ability to repurpose distributed capacity, and participation in shared infrastructure. Those mechanisms can matter to a large technology operation, but the sources do not quantify their net financial value for PayPal.
No independent study in these accounts isolates a current, causal company-wide financial return attributable to PayPal’s Linux adoption. The 2007 Linux economics account is historical; the OpenStack availability number is a stated requirement; and the Kafka scale and HERA latency figures are claims in PayPal-authored material. They support a historical case study, not a current savings estimate or a universal ranking of open-source and proprietary systems.
For another organization, the relevant comparison would include licensing and infrastructure costs, the expense of running and securing distributed systems, the value of consistent test environments, and the effort required to contribute upstream or maintain internal changes. PayPal’s accounts show why it considered those trade-offs worthwhile in particular systems, but do not settle them for every company.
Quick Recap
Sources
- LinuxInsider: PayPal Uses Linux, Open Source to Support Growth (October 19, 2007), a first-person account by Matthew Mengerink.
- Open Infrastructure Foundation: PayPal OpenStack case study (May 7, 2014).
- PayPal Developer Community: Kafka at PayPal (September 7, 2023).
- PayPal Technology Blog: HERA: A Whole New World of Database Access (July 23, 2019).
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
Do these 3 things before closing this tab:
1Repair Windows errors before they cause bigger problems2Scan for outdated or missing drivers - takes under a minute3Clear out junk files and repair common Windows errors
