|
|
» Learn » How It's Better
Why UniCluster 4.1?
Below are some of the differentiators that make UniCluster 4.1 the superior alternative for integrated HPC cluster management.
To see how UniCluster 4.1 compares to alternatives, review the comparison chart.
- HPC Systems Management capabilities
Only UniCluster 4.1 applies a systems management approach to cluster operation, addressing several key phases of the HPC lifecycle.
- Key applications and tools packaged as Kits
UniCluster 4.1 includes numerous technology packages options to extend the power of your base cluster, such as Nagios, Cacti, MPI Libraries, SGE, OFED, Virtualization Kit, and an Optional Intel Cluster Checker Kit.
- Virtualization capabilities (Beta version)
UniCluster 4.1 offers Xen node provisioning, with the ability to create nodes running Linux with Xen Kernels and install images onto nodes running Xen.
- Community-enabled open source
With UniCluster, users gain access to a vibrant community of users for knowledge sharing and support.
- Unrestricted support
Unlike other products, UniCluster is fully supported – every component of the product has commercial-grade support available.
- Easy to acquire and implement
UniCluster installation is fast and easy, with a simple, seamless upgrade process. And with support via the both Grid.org community and Univa, users can rest assured their implementations are sustainable.
Compare UniCluster 4.1 capabilities to alternative cluster software solutions:
| Product |
Platform Computing Platform LSF |
Altair Engineering PBS Pro |
Univa UniCluster 4.1 |
| Licensing |
| Price Model |
Per CPU |
Per Core* |
Per Node & CPU Hour* |
| Price Structure |
Perpetual and Annual Subscription |
Annual per core. |
Annual Subscription and Pay-per-Use* |
| Class Based Pricing |
Yes |
No |
No |
| Memory Restrictions |
Yes |
No |
No |
| Core Surcharge |
Yes |
No |
No |
| Software License |
Closed |
Closed |
Open |
| License Key Required |
Yes |
Yes |
No |
| Bundled Components |
| Workload Scheduler |
Platform LSF |
PBS Pro |
Sun Grid Engine |
| Cluster Management |
Extra* |
N/A |
Included |
| Realtime Monitoring |
Extra* |
N/A |
Included |
| MPI |
Extra* |
N/A |
Yes |
| Cluster Tools |
No |
No |
Yes |
| Bundled Operating System |
No |
No |
Yes |
| Oracle Unbreakable Linux |
N/A |
N/A |
Yes |
| CentOS |
with OCS |
N/A |
Yes |
| RHEL |
with OCS |
N/A |
Yes |
| Support |
| Multiple Levels (price points) |
24 Hour Only |
No |
Yes |
| Single Support Path |
No |
No |
Yes |
| * for bundled components |
|
|
|
| Cloud Functionality |
| Cloud Extension |
No |
No |
Yes |
| Provisioning in Cloud |
No |
No |
Yes |
| Price Model |
N/A |
N/A |
Pay-per-Use |
| Intel Cluster Ready Compatible Component |
No |
Yes |
Yes |
| Cluster Management Functionality |
| Linux/Windows Dual Boot Support |
No |
No |
Yes |
| Consistent, Single Deployment Mechanism |
No Consistency* |
N/A |
Yes |
| Single User Sign-on |
No |
No |
Yes |
| Bare Metal Provisioning |
No |
No |
Yes |
| Configuration Management |
No |
No |
Yes |
| Manages Application Install into Cluster |
No |
No |
Yes |
| Native Package Management |
No |
No |
Yes |
| Integrated Monitoring |
LSF information only |
No |
Yes |
| Grid Services |
Extra* |
N/A |
Included |
| Workload Management Functionality |
| Scheduling Policies |
Fairshare, Preemption, Application Encapsulation, Advance Reservation, Resource Reservation and Service Level Agreement (SLA) based scheduling |
Fairshare, Preemption, Advance Reservation, Backfill, Resource Reservation, Load Balancing |
Fairshare, Functional Project, Override, Advance Reservation, Backfill, Resource Reservation and Resource Quota Sets |
| Checkpointing |
user, application and kernel |
application or kernel |
Flexible checkpointing method applied using any technique - user library or kernel. |
Parallel Jobs Scalability
* Note this describes scalability for the workload manager only.
|
scalable to 6,000 way parallel job |
single parallel job with 10,000 cores |
single parallel job with 64,000 cores |
| Node Scalability
* Note this describes scalability for the workload manager only.
|
Scalable to 4,000+ nodes |
3,000+ nodes |
Scalable to 4,000+ nodes 64,000 cores |
| Standards |
DRMAA, JSDL |
DRMAA, POSIX |
DRMAA, POSIX |
| Job Arrays |
Yes |
Yes |
Yes |
| Resource Quotas |
Partial with general limits |
No |
Resource Quota Sets |
| Interactive Job Support |
full interactive support |
full interactive support |
full interactive support |
| Notes: |
| |
Platform licenses Cluster Management, Realtime Monitoring, MPI and Grid Services at extra cost. See www.platform.com (Platform Manage; Platform OCS, Platform RTM, Platform MPI, Platform LSF MultiCluster) |
Altair Engiineering offers PBS Pro licenses from their online store. Altair does not offer cluster management, cluster tools or monitoring. |
Univa offers CPU Hour & pay-per-use pricing via UniCloud. UniCluster pricing is per node. |
| |
Platform products have independent installers. Platform OCS and Platform Manage each have discreet package maagement and installers. |
|
UniCluster leverages a single package management and installer framework. Individual applications can also be installed as RPMs. |
Information may change and is provided for informational purposes. All trademarks are the property of their respective owners.
We would be pleased to correct any mistakes. Please contact us directly at 1-800-370-5320.
|
Schedule LIVE Demo
See UniCloud in a range of live scenarios
Find out more »
Oracle Unbreakable Linux Support
Now available with UniCluster 4.1

Find out more »
|