DumCan open source software cope with petabytes of data? We’re about to find out.
Administrators at the European Center for Nuclear Research (CERN), after moving in lockstep with Microsoft longer than their researchers have been smashing atoms with the Large Hadron Collider, appear to have reached the breaking point with the giant technology company.
Rather than pay their vendor of long standing – Microsoft – a 10x increase to license mission-critical software, they’re developing their own.
In a blog posted last week, Emmanuel Ormancy, a systems architect at the Geneva-based laboratory, said CERN’s Microsoft Alternatives project will take an open-source approach to wresting control of core functionalities and data from the software company. The project, known internally as MAlt, has run since 2018 and will trot out a pair of pilot platforms later this year.
The reason: vendor lock-in, which occurs when users become over-reliant on providers of the proprietary products and support services needed to run their operations.
According to Ormancy’s postOpens a new window , the Redmond, Washington-based company leveraged its pervasive reach across the breadth of CERNs operations to extract premium pricing for software and services in recent contract negotiations.
Rather than continuing the subsidies accorded academic institutions, which CERN enjoyed for the first two decades of its relationship with Microsoft, the new contract subjects it to a pricing schedule based on usage more akin to what commercial entities pay for software and support.
Collision Course
Despite a series of graduated raises built into the deal over the new contract’s 10-year life, Ormancy describes CERN’s situation as unsustainable for an organization that mostly relies on vendors for computing and storage equipment. Never mind a research lab that generates as much data in a second as some commercial firms in a yearOpens a new window when it slams sub-atomic particles into one another at the speed of light.
Hence, 2,500 research professionals and staff and 12,000 adjunct users will migrate to IT platforms that lessen the institute’s dependency on commercially-produced software.
CERN maintains agreements with more than 40 vendorsOpens a new window , including Oracle’s AutoVue collaboration kit and the Code Composer Studio that Texas Instruments publishes to help customers manage its silicon.
The laboratory licenses Microsoft’s Windows operating system as well as its Office applications and its Visual Studio software development kit. Those licenses, paid for by taxpayers in the 23 member countries that fund the lab, cover office and home PCs for lab professionals and administrators, according to the CERN website.
The first casualties, Ormancy wrote, are Microsoft’s Outlook and Skype for Business services. The pilots of replacement platforms that begin in September are the initial steps in the initiative that could see more of the lab’s IT landscape given over to open-source software.
Making the project public is part of CERN’s drive to drum up support among institutions that it says are vulnerable to similar risks when vendor lock-in translates to pricing power. While tech vendors are eager to cultivate a do-gooder approach when doling out subsidized software for research, there’s no denying their bottom-line focus.
CERN’s is a strength-in-numbers appeal, given that more than 100 research facilities and universities share data and analysis from the lab. And it is a key elementOpens a new window because a wide body of contributors ensures that any move to open-source achieves its cost-containment objective.
Hiding in Plain Sight
It’s an ambitious exercise in light of the infrastructure requirements for collecting, parsing and storing the massive amounts of data generated in the collider’s four detection chambers. When in operation, the super collider produces 600 million collisions per second, each of which generates around a million bytes of data.
Analytics engines then cull those flows with algorithms in a two-stage processOpens a new window that lets researchers select 200 or so interesting events. The events then are catalogued in a tape archive that surpassed 200 petabytes in July 2017. Under construction for a decade, the collider’s 16.5-mile tunnel buried 575 feet below ground on the Swiss-French border came online in 2008.
While CERN’s open-source news appeared to take Microsoft by surprise – the company declined comment for several days after Ormancy’s post and provided boilerplateOpens a new window when it finally did respond – the lab had foreshadowed its intentions under the nose of the vendor.
In May, CERN hired an outside consultant to explore the virtues of open-source firmware for managing server networks, like the 230,000 processor cores that are bundled in its data centers.
It also maintains more than 50 code repositoriesOpens a new window on the GitHub open-source development platform that Microsoft bought last year.