Logger Script


cloud HPC

'cloud HPC (High Performance Computing) service is for providing clusterized ucloud server and solution for high-performance operation.

What is cloud HPC?

1. Researchers or engineers can solve high bandwidth, network with short delay time, complex scientific and engineering analyses and interpretation simulation in using applications requiring high-capacity computing power and business-related problems.

2. In general, users should wait for a long time to use shared cluster resources or purchase expensive hardware system themselves.

HPC service provides multiple virtual machines by binding into one cluster for mass calculation or data saving. This cluster is composed with calculation node which is virtual machine providing computing resource and head node, an administrator connecting calculation nodes to make them seen as one single system. Users submit jobs to solve problems on head node and head node schedules necessary resources based on priority and current usage of calculation node and distributes jobs on calculation nodes. You can receive reporting on calculated results through head node. Compute Manager helps to control this series of operation process through GUI.

Explanation on terms
- Head node : Computer for HPC control executing management function on multiple calculation nodes
- Calculation node : Computer for calculation executing mass operation under head node's control
- PBS Professional : Software solution managing HPC workload
- Compute manager : Web service enabling execution of instruction and file transmission with simple web browser operation that used to be executed with conventional telnet and FTP

Introduction of Solution

PBS Professional짰was developed by Ames institute of NASA in the early 1990s and it is being used in world TOP 500 large-scale HPC(High Performance Computing) systems as its security is well acknowledged. PBS Professional not only manages HPC resources including CPU, memory, disk, network, software license etc. to increase efficiency of HPC system but suggests optimum HPC resource utilization plans through analysis on operational environment.

Compute Manager of Altair is easy-to-use, web browser-based HPC interface. You can easily execute operations with utilizing HPC, monitor status of running operations and check results from the progress with chart and animation in real time.
All solving operations utilizing HPC can be simply executed through Compute Manager and engineers do not have to agonize anymore about how to configure and execute MPI (massively parallel processing) for large-scale simulation operations.

Service Highlights

1. Economics
Customers can compose and operate HPC cluster by paying hourly or monthly rate as they want.

2. Extendability
Server can be added and removed according to size and time necessary for workload as you want.

3. Efficiency
Working process can be done innovatively fast as you can use computing resource only in a few minutes
after computing operation starts with management tool (GUI) without waiting for a while.

4. Security
PBS Professional, the management tool for entire operation executed on HPC, enhanced reliability
on security function with solely obtaining EAL 3+ level security authentication in entire industry.

Major functions of service

1. PBS Professional : Guarantee on the industry's highest HPC safety and increase in utilization
- Providing optimum scheduling algorithm to increase resource utilization
- Extendability verified through large-scale system supply to NASA, Meteorological Administration and Pentagon
- Providing functions to detect failure on calculation node and to automatically correct user error (hook) to minimize operation failures
- Security enhancement through solely obtaining EAL(international standard security authentication) 3+ level in entire industry

2. Compute Manager (Easy to use) : Increase in convenience of HPC user
- Providing WEB-based HPC environment
- Function to monitor execution results in real time and to edit file
- Enabling teams to cooperate and removing repetitive operations with profile management of HPC execution environment

3. Upgrade and technical support (Keep Track and Plan)
- Guaranteeing continuous addition of function and improvement of performance
   * Hook / Integration of MPI library / Green Computing (Function of power saving)
       * Improvement on Job execution speed (7 times) / Improvement on re-operation speed (40 times) / Expansion of support
          performance for concurrent access (15 times)
-Global technical support available at 27 offices in 14 countries from Altair Engineering Inc.
- Operating 24-hour, WEB and Knowledge-Based online community
- Premium supporting from partner companies and Altair Korea for failure support and customer reception

Physical Server Computing

Classification Physical Server-based Computing cloud HPC
Extendability Limited physical capacity on an existing system
Addition/Replacement of Hardware required for expanding capacity
Quick expansion of HPC cluster available by adding computing node
Promptness Installation time such as Delivery and Installation required
Time for Hardware Maintenance required
Integrated installation of OS and HPC scheduler
Automatic configuration of NFS, NIS, SSH etc. for composing Cluster
Stability Out of service with Hardware errors
System Redundancy required in case of errors.
Switch to spare Cloud resources, Prompt recovery from errors
System Redundancy unnecessary
Cost Reduction Purchasing cost for Hardware including spare capacity Operation costs such as
Network/Field/Electricity incurred when terminating a period necessary
Use of capacity as much as needed Charged by usage
Reduction on additional operation costs

Expected Effect

1. Increase in business efficiency
- Users can focus only on their work without knowledge for use of HPC system. (WEB-based UI provided)
- Repetitive operations reduced by minimizing HPC operation failures
- Reducing HPC work time with optimization of execution environment on each Application
- Smooth cooperation feasible between members through Profile sharing

2. Maximization of HPC resource utilization
- Able to use HPC resources on weekends or holidays with Queue Reservation functions
- Minimization of operation failures through dualization, Job correction function, automatic detection for failing equipment
- Able to improve effectiveness gained by investment through analysis on resource record and systematically calculate adoption cost

Service Rate

1. Rate of Use in Basic Service : server + PBS Professional
(Unit:(USD)/ month,hour)
Classification Specification included in Default server PBS Professional Total
Monthly Rate
Monthly Rate
Monthly Rate
4 4 100 $95 $0.132 $43 $0.065 $138 $0.197
4 8 $151 $0.210 $43 $0.065 $194 $0.275
8 8 $191 $0.265 $43 $0.065 $234 $0.330
8 16 $300 $0.418 $43 $0.065 $343 $0.483
12 16 $340 $0.473 $43 $0.065 $383 $0.538

* Default OS : CentOS 5.8 (64Bit)
* PBS Professional charges calculation node only.

2. Additional SW (optional)

Computer manager

- Monthly Rate(user): $123
- Hourly Rate(1h/user): $0.201
3. Network rate system(Same as server)
[Over 16GB memory server : Network rate]
Monthly usage less than 2 TB per cloud server
(over 16GB memory VM)
Monthly usage by customer (on an ID basis) over 2 TB Refer to the table below

(Unit : (USD)/GB)
Classification Usage Section Application by Cloud server Application by Customer (by ID)
under 2 TB
per Cloud server
(over 16GB memory VM)
2TB ~ 10TB 10TB ~ 20TB Over 20TB
Meter-rate Unit cost by section
(per GB)
Free $0.082 $0.073 $0.064
Flat Rate
(monthly unit)
30 TB/month $1,802
50 TB/month $2,703

[less than 16GB memory VM : Network rate]
Monthly usage less than 1 TB per cloud server
(less than 16GB memory)
Monthly usage by customer (on an ID basis) over 1 TB Refer to the table below

(Unit : (USD)/GB)
Classification Usage Section Application by Cloud server Application by Customer (by ID)
under 1 TB
per Cloud server
(less than 16GB memory)
1TB ~ 10TB 10TB ~ 20TB Over 20TB
Meter-rate Unit cost by section
(per GB)
Free $0.082 $0.073 $0.064
Flat Rate
(monthly unit)
30 TB/month $1,802
50 TB/month $2,703

4. Additional Disk and Public IP
Classification Unit Monthly Rate((USD)/month) Hourly Rate((USD)/hour)
Additional Disk per add 10GB
(300 GB at maximum)
$0.631 $0.001
Additional Public IP per 1 IP $5 $0.007

Virtual Machine Alert
Option 1.
  • Select Server (CPU/RAM)
  • : Server(VM) 1v core/1G memory
  • Operatoion System
  • :
  • Comment
  • :