PDF version of this report
You must have Adobe Acrobat reader to view, save, or print PDF files. The
reader is available for free
download.
Altair Univa
Grid Engine
Copyright 2020, Faulkner Information Services. All
Rights Reserved.
Docid: 00018700
Publication Date: 2011
Report Type: PRODUCT
Preview
The Univa Grid Engine provides software for resource and workload management
in addition to dynamic resource provisioning across the datacenter.
Originally part of Oracle’s line-up after its purchase of Sun Microsystems
in 2010, Oracle transitioned the product line and support to Univa in the
fall of 2013. In September 2020, Univa was acquired by Altair, a provider of
HPC workload and workflow management technology. Grid Engine is a scalable
workload management system and its software can link across networked
server, workstation, and desktop environments to help establish a common
grid for the better allocation of processes.
Report Contents:
- Description
- Related Faulkner Reports
- Vendor
- Applications
- Environment
- Support
- Pricing
- Competitors
- Web Links
Description
[return to top of this
report]
The Altair Univa Grid Engine, originally part of Oracle’s software lines
after its purchase of Sun Microsystems, is a scalable workload management
system. Oracle transitioned sales and support to Univa in the fall of
2013. Altair acquired Univa in September 2020.
Related Faulkner Reports |
Oracle Company Profile |
Oracle Cloud Computing Services Product Profile |
The Grid Engine is touted as the industry’s most widely deployed scalable
workload management system.
Grid Engine software is a distributed resource management (DRM) system
that manages the data center distribution of user workloads to available
compute resources. Recent updates have added cloud computing support.
The modern version of Grid Engine began as the updated commercial
revision of the open-source N1 Grid Engine project.
Vendor |
Name: Univa Corporation |
By distributing workloads to the most appropriate available resources,
Univa states that the Grid Engine can extend utilization rates by as much
as 90 to 95 percent. Traditional data center compute resources offer just
10 to 25 percent increases.
When users submit their work to Altair Univa Grid Engine as “jobs,” the
software monitors the current state of all resources in the cluster and
assigns these jobs to the best-suited resources. Grid Engine offers the
flexibility to model a computing environment as resources and to translate
business rules into policies that govern the use of those resources.
In 2018, along with the release of version 8.6.0 of Grid Engine, Univa
boosted performance of the data store, the memory allocator, and the
scheduler. In addition, the communication library was improved to support
data compression before transferring data over the network. Currently, the
Grid Engine is in version 8.6.17, which was released on October 30, 2020.
It offers support for Docker jobs’ parameters and compatibility with DCGM
versions up to 2.0.10.
Perhaps Grid Engine’s greatest feature is its ability to scale to large
clusters, up to 200,000 in a single managed environment. It can manage a
grid of up to 10,000 nodes and run more than 200 million jobs on a monthly
basis. Its scalability is enabled through its service domain manager,
which migrates resources among Grid Engine clusters to accommodate
workload. Other benefits of Grid Engine include:
- Native Docker Support – Enables the running of Docker
containers in a Grid Engine cluster at scale to blend containers with
other workloads. - Support for GPU – Allows GPU frameworks to scale from
containers and servers to clusters and clouds. - Reporting – Monitoring features track and measure
resource utilization in workload management clusters. - HPC Support – HPC workloads and microservices fun on
shared Kubernetes clusters.
Other Grid Engine features are detailed in Table 1.
Feature | Description |
---|---|
Priority and Utilization Policies | Delivers multiple scheduling policies for matching workload in the cluster to key business objectives like reducing job turnaround time and workload prioritization. |
Scalability | Scales to a cluster of up to 200,000 cores in a single, on-premise environment. |
Resource Management | Collects metrics from all cluster nodes and uses this information to evaluate workloads that are pending to match them to available resources. |
Multiple Workloads | Accepts any application or workload accelerator including Docker, Intel KNL, and NVIDIA GPUs. |
Quotas and Limits | Configures flexible quotas on users, projects, and groups in an effort to control how much workload is run in the cluster. |
Add-on Feature | License Orchestrator – Manages the allocation of licensed applications across Grid Engine clusters. |
Applications
[return to top of this
report]
The Grid Engine is suitable for computing uses from small scale to Big
Data. It offers administrators the flexibility to model computing
environments as resources to translate business rules into policies that
govern the use of those resources. It can create computing grids at the
department, multi-department, and enterprise level and can also
incorporate and integrate thousands of Windows desktops, servers, and
other platforms into grids.
Environment
[return to top of this
report]
Grid Engine is interoperable with open network products.
Table 2 summarizes Oracle Grid Engine’s platform support.
Operating System | Version | Architecture |
---|---|---|
Microsoft Windows | 10 (Pro and Enterprise), 8 (8, 8.1 Pro, and Enterprise), Server 2012 (Server 2012 and Server 2012 R2), Vista (Enterprise and Ultimate), 7 (Professional, Enterprise, and Ultimate), Server 2008 (Server 2008 and Server 2008 R2) |
32-bit, 64-bit |
Microsoft Windows | Server 2003 (Server 2003 and Server 2003 R2); XP Professional (SP3) |
32-bit |
Oracle Solaris | 10, 11 | x86-64, SPARC 64-bit |
Apple macOS | 10.11 or higher | x86, x86-64 |
Oracle Linux | 6 or higher, 7 | x86, x86-64 |
Additional operating systems supported include SLES, RHEL, CentOS,
openSUSE Leap, Ubuntu, HP-UX, and IBM AIX.
Support
Univa provides product support for Univa as well as Open Source Grid
Engine 6.2U5 and variants of Open Grid Scheduler and Son of Grid Engine
8.1.9. Support is available through https://univa.zendesk.com/
or 800-370-5320.
Pricing
[return to top of this
report]
Grid Engine can be purchased from Altair. Interested parties can also
download a free trial of Grid Engine by completing this form.
Competitors
[return to top of this
report]
The Grid Engine faces competition from IBM Spectrum Computing, the Globus
Toolkit, and the World Community Grid.
Web Links
[return to top of this
report]
- IBM Spectrum Computing: http://www.ibm.com/it-infrastructure/spectrum-computing/
- Oracle: https://www.oracle.com/
- World Community Grid: https://www.worldcommunitygrid.org/
[return to top of this
report]