Visit additional Tabor Communication Publications
January 18, 2011
Univa announced today it would be acquiring the Sun/Oracle Grid Engine engineering expertise from Oracle Corp. In doing so, the company will take over stewardship of the popular open source workload manager, which, in the space of two years, has passed through three companies: Sun Microsystems, Oracle, and now Univa. Its new owners plan to support existing deployments of Grid Engine as well as develop a commercial version with added capabilities.
Although Univa is not talking publicly about the terms of the agreement (if any) that enabled them to spirit away the Grid Engine team from Oracle, Tuesday's press release did specify that the principal engineers from Grid Engine team, including Grid Engine founder and original project owner Fritz Ferstl would be joining Univa. To sweeten the deal, Ferstl has been named CTO and will head the company's EMEA business.
By jettisoning the Grid Engine team, Oracle is continuing to shed its associations with high performance computing, and with open source software, in general. Although Grid Engine is a general-purpose workload manager, and has many applications outside of HPC (including within Oracle's own Enterprise Manager suite), the company is apparently not inclined to be the steward of this technology. At the very least, it is unwilling to develop and maintain the open source software for the larger community -- that despite this gushing endorsement of Grid Engine that Oracle posted in September 2010 on its website:
"Following the acquisition, Sun Grid Engine came to Oracle Enterprise Manager and we are absolutely delighted to welcome OGE to the Enterprise Manager family. The product is absolutely fantastic and a leader in its market space. We plan to aggressively support and market this product."
Oracle's original plan was to "enterprise up" the Grid Engine in order to make it more suitable for the company's business-centered customer base. But apparently if someone else is willing to do the heavy lifting with regard to maintaining the foundational software, so much the better.
In passing the torch to Univa, Oracle is almost certainly doing the HPC community a big favor. The concern with Oracle was its antipathy in pursuing an HPC path for the technology. This would eventually have forced Grid Engine to fork into one or more commercial versions and possibly even multiple open source versions. This could still happen, but given Univa's large base of HPC customers, its own stake in the software, and its experience with open source development, this is much less likely to occur.
According to Univa CEO Gary Tyreman, their plan is to keep all the Grid Engine users happy by ensuring everyone has a path forward. This includes those who elect to stay with the open source version (mainly government research labs and academic types), commercial users who need Grid Engine support but are not ready to upgrade, and those customers looking for a commercially-supported product and ongoing evolution of the software.
It also includes Univa itself, who currently employs the workload manager in its UniCluster offering. In addition, the company will rely on Grid Engine in its UniCloud product to manage workloads across distributed resources. As of late, the company is shying away from the cloud terminology, and replacing it with a "datacenter optimization" mantra that dovetails nicely with how Univa intends to use Grid Engine across its product portfolio.
The responsibility for Grid Engine also brings with it some strange bedfellows. For example, Univa is now in the position of supporting competing offerings, such as a Grid Engine-equipped Rocks Cluster distribution. "We are redrawing the competitive dynamics, not just for the industry, but in particular for ourselves," notes Tyreman. "Overnight some of my competitors become my best friends."
The overarching goal, he says, is to unify the community "as it was." Tyreman insists they are committed to working with the Open Grid Scheduler (OGS) project, which will be the repository for the open source version of the software. "There will always be an open source version of Grid Engine available," he declares.
Tyreman realizes that the "free" version will be sufficient for a large portion of the HPC market. That include some of the biggest single-system deployments, such as the Ranger supercomputer at the Texas Advanced Computing Center and the TSUBAME machine at the Tokyo Institute of Technology. But in general, the industry is moving toward a two-tiered model with regard to open source. Just as Platform LSF and PBS Pro have their open source counterparts in Platform Lava and OpenPBS/Torque, respectively, Univa believes that there is plenty of demand for a commercial Grid Engine offering.
According to Tyreman, there are probably 4 million-plus CPUs spread out across more than 1,000 organizations running Grid Engine today. Univa only needs a fraction of that market to make a commercial product economically viable. Businesses with high-value applications such as electronic design automation (EDA), manufacturing design, and financial analytics would be among those most interested in a shrink-wrapped premium offering.
Like other commercial workload schedulers, Univa's Grid Engine will be licensed on a per-core, per-term basis. In general, Tyreman expects licensing costs to be aligned with traditional Grid Engine-based products, and to be very competitive with rival management products.
The real balancing act for Univa will be to differentiate their own product, while keeping the open source base viable for the larger community upon which they depend. Although the value-add product that Univa plans to license will certainly be for technical computing customers looking for extra features and functionality, what exact shape that will take remains to be seen. With the advent of enterprise cloud computing, and Univa's own interest in that space (although lately the company has been shying away from the cloud terminology), it's likely that they will evolve their product to serve a broader set of customers than just HPC users. Workload scheduling is one thing in a 100-node HPC cluster, and quite another in a datacenter with thousands of servers running mission critical applications.
Grid Engine end users and companies are probably breathing a little easier now that the software is back in the hands of a genuine HPC vendor that understands the ecosystem. After its Oracle detour, the Grid Engine roadmap appears to be returning to the one established by its founders. Says Tyreman: "We are going to continue on a path that was already set 10 years ago."
May 16, 2013 |
When it comes to cloud, long distances mean unacceptably high latencies. Researchers from the University of Bonn in Germany examined those latency issues of doing CFD modeling in the cloud by utilizing a common CFD and its utilization in HPC instance types including both CPU and GPU cores of Amazon EC2.
May 15, 2013 |
Supercomputers at the Department of Energy’s National Energy Research Scientific Computing Center (NERSC) have worked on important computational problems such as collapse of the atomic state, the optimization of chemical catalysts, and now modeling popping bubbles.
May 10, 2013 |
Program provides cash awards up to $10,000 for the best open-source end-user applications deployed on 100G network.
May 09, 2013 |
The Japanese government has revealed its plans to best its previous K Computer efforts with what they hope will be the first exascale system...
May 08, 2013 |
For engineers looking to leverage high-performance computing, the accessibility of a cloud-based approach is a powerful draw, but there are costs that may not be readily apparent.
05/10/2013 | Cleversafe, Cray, DDN, NetApp, & Panasas | From Wall Street to Hollywood, drug discovery to homeland security, companies and organizations of all sizes and stripes are coming face to face with the challenges – and opportunities – afforded by Big Data. Before anyone can utilize these extraordinary data repositories, however, they must first harness and manage their data stores, and do so utilizing technologies that underscore affordability, security, and scalability.
04/15/2013 | Bull | “50% of HPC users say their largest jobs scale to 120 cores or less.” How about yours? Are your codes ready to take advantage of today’s and tomorrow’s ultra-parallel HPC systems? Download this White Paper by Analysts Intersect360 Research to see what Bull and Intel’s Center for Excellence in Parallel Programming can do for your codes.
In this demonstration of SGI DMF ZeroWatt disk solution, Dr. Eng Lim Goh, SGI CTO, discusses a function of SGI DMF software to reduce costs and power consumption in an exascale (Big Data) storage datacenter.
The Cray CS300-AC cluster supercomputer offers energy efficient, air-cooled design based on modular, industry-standard platforms featuring the latest processor and network technologies and a wide range of datacenter cooling requirements.