https://wiki.egi.eu/w/api.php?action=feedcontributions&user=Dmg&feedformat=atomEGIWiki - User contributions [en]2024-03-29T10:51:11ZUser contributionsMediaWiki 1.37.1https://wiki.egi.eu/w/index.php?title=VT_EGI_Pay-for-Use_Service/Price_Overview&diff=72713VT EGI Pay-for-Use Service/Price Overview2014-12-16T20:21:02Z<p>Dmg: /* Resource Providers, Services and Prices */</p>
<hr />
<div>{{Template:EGI-PayForUse-PoC_menubar}} {{TOC right}} <br />
<br />
The following table provides an overview of the technical details of the resource and technology providers for the PoC. <br />
<br />
Further below are instructions for adding pricing into the accounting mechanisms. <br />
<br />
= Resource Providers, Services and Prices =<br />
<br />
{| class="wikitable"<br />
|-<br />
! scope="col" | NGI/RC <br />
! scope="col" | GOCDB Name <br />
! scope="col" | Contact <br />
! scope="col" | Service Type <br />
! scope="col" | Services and Resources <br />
! scope="col" | Preferred Payment Direct/Indirect <br />
! scope="col" | Grid Compute Price<br>[HEPSPEC06/hr] <br />
! scope="col" | Cloud Compute Price<br>[wallclock/hr] <br />
! scope="col" | Grid Storage Price<br>[€/GB/Month] <br />
! scope="col" | Cloud Storage Price<br>[€/GB/Month] <br />
! scope="col" | VAT<br>[%]<br />
|-<br />
! scope="row" | Albert Einstein Center, Univ. of Bern <br />
| UNIBE-LHEP <br />
| Sigve Haug <br />
| Batch via CENTOS 5/6 ARC CE <br> <br />
| <br />
*500 cores (approx. 2 months in 2013) on highest priority; <br />
*Consumption preferably in one go, but distributed in time also possible <br />
*10-20 TB disk on local SE for consumption period via DPM SRM interface plus two weeks<br />
<br />
| Both <br />
| €0.03/hr <br />
| N/A <br />
| No long-tem storage offered. Disk up to 20 TB for CPU usage period plus to weeks for free. <br />
| <br />
| 8%<br />
|-<br />
! scope="row" | NGI_BG <br />
| BG01-IPP<br />
| Todor Gurov<br>Emanouil Atanassov <br />
| <br />
Batch &amp; Grid: EMI-2/EMI-3<br />
<br />
| <br />
*16 NVIDIA Tesla M2090 6 GB + 48 core Intel Xeon E5679 @ 2.53GHz with 4 GB of RAM per core <br />
*576 cores Intel Xeon X5560 @ 2.8GHz with 1.5 GB of RAM per core <br />
*MPI jobs <br />
*Non-blocking InfiniBand interconnect <br />
*132 TB of storage provisioned through Lustre FS (/home and /scratch systems), EMI dCache<br />
<br />
| Both <br />
| <br />
*€1.00/GPU/hr. + VAT (GPU-enabled servers) <br />
*€0.02/core/HEPSPEC06/hr. + VAT (CPU only Intel Xeon X5560-based cluster) <br />
<br />
| N/A <br />
| €0.04/GB/month (+VAT) via dCache. Maximal capacity 30TB. <br />
| <br />
| 20%<br />
|-<br />
! scope="row" | CESGA-IBERGRID <br />
| CESGA <br />
| Ivan Diaz <br> Ruben Diez <br />
| Batch + Cloud <br />
| <br />
*720 cores (Batch) AMD Opteron(tm) Processor 6174 2.2Ghz <br />
*240 cores (Cloud) Intel(R) Xeon(R) CPU E5520 2.27GHz <br />
*MPI jobs <br />
*RAM from 1GB to 32GB per core<br />
<br />
| Direct <br />
| <br />
€0.02/core/HEPSPEC06 hour <br />
<br />
| €0.04/core+2GB RAM/hour (+VAT) Example: <br />
*small: 1 core &amp; 2GB €0.04/hour <br />
*medium: 2 cores &amp; 4GB €0.08/hour <br />
*large: 4 cores &amp; 8GB €0.16/hour<br />
<br />
| €0.055/GB/month (+VAT) <br />
| €0.055/GB/month (+VAT) <br />
| 21%<br />
|-<br />
! scope="row" | IFCA-IBERGRID <br />
| IFCA-LCG2 <br />
| Jesus Marco <br />
| Batch (SLURM, SGE) + Cloud <br />
| <br />
*Batch via CREAM CE <br />
*Cloud via OpenStack <br />
*Up to 2500 cores in servers with 4, 8, 16, 40 physical cores <br />
*Infiniband access to other nodes and storage possible for MPI jobs <br />
*RAM from 4GB to 256GB per core (max 1TB)<br />
<br />
| Both, Direct preferred <br />
| <br />
*€0.05/core/hour for usual instances,&nbsp;&lt;0.01/core/HS06 hour (+ VAT) <br />
*€0.04/core/hour for multicore use (&gt;128 cores)<br />
<br />
| €0.03 <br />
| Minimal storage free, requests up to 100 TB high performance storage possible <br />
| <br />
| 21%<br />
|-<br />
! scope="row" | Institute of Informatics of the Slovak Academy of Sciences (II SAS) <br />
| IISAS-FedCloud <br />
| <br />
Ladislav Hluchý<br> Viet Tran <br />
<br />
| Cloud (Openstack) <br />
| <br />
*120 cores (Cloud) Intel(R) Xeon(R) CPU E5570 2.93GHz <br />
*1, 2, 4, 8 cores VMs <br />
*RAM up to 4GB per core<br />
<br />
| Both <br />
| N/A <br />
| €0.06/core/h +VAT <br />
| TBD <br />
| <br />
| 20%<br />
|-<br />
! scope="row" | Latvian Grid <br />
| IMCSUL <br />
| Kaspars Krampis <br />
| Cloud: OpenStack, GRID: ARC + SLURM <br />
| <br />
*CPU: Up to 256 cores, RAM up to 4GB per core<br />
<br />
| TBC <br />
| €0.07/core/hr + VAT <br />
| N/A <br />
| TBD <br />
| <br />
| 21%<br />
|-<br />
! scope="row" | MASTER-UP Srl <br />
| UNI-PERUGIA <br />
| <br />
Alessandro Costantini<br>Antonio Lagana <br />
<br />
| EMI1/2, CEs, WNs, SE <br />
| <br />
*EMI1 middleware (phasing to EMI2) - CE's, WN's and SE. <br />
*10 nodes to start (intel 4core, ethernet connection) <br />
*Long experience in Molecular Science and Comp Chem, can provide support for such innovative applications and in the field of R&amp;D for combustion, energy and material science. <br />
*Managing a computer farm of 200 cores and 4TB of storage part of those resources are supporting EGI.<br />
<br />
| Direct <br />
| €0.05 <br />
| N/A <br />
| €0.01/GB per month <br />
| <br />
| 22%<br />
|-<br />
! scope="row" | NGI_GRNET <br />
| All HG-* <br />
| Kostas Koumantaros <br />
| Cloud and Grid <br />
| <br />
*TBC<br />
<br />
| TBC <br />
| €0.05/hr + VAT <br />
| €0.05 + VAT <br />
| TBD <br />
| <br />
| 23%<br />
|-<br />
! scope="row" | NGI-IT <br />
| <br />
INFN-BARI PRISMA-INFN-BARI (cloud) <br />
<br />
INFN-XX <br />
<br />
| Vincenzo Spinoso <br />
| EMI, OpenStack <br />
| <br />
*gLite MW stack: Cream CE, LFC, Storm SE, WMS+LB, BDII. <br />
*Cloud: OpenSTACK<br />
<br />
| Both <br />
| €0.07/hr + VAT <br />
| €0.05 + VAT <br />
| €0.02 + VAT <br />
| <br />
| 22%<br />
|-<br />
! scope="row" | 100 Percent IT <br />
| 100IT <br />
| David Blundell <br />
| Cloud (Openstack) <br />
| <br />
*Intel Xeon X5670 @ 2.93GHz <br />
*Cloud via OpenStack<br />
<br />
| Both <br />
| N/A <br />
| €0.07 + VAT <br />
| TBD <br />
| <br />
| 20%<br />
|-<br />
! scope="row" | PL-Grid <br />
| CYFRONET-LCG2 <br />
| Mariusz Sterzel <br />
| Cloud + Batch via Cream / EMI <br />
| <br />
*Computing power, storage, and human support <br />
*Batch processing via Cream CE/EMI <br />
*Cloud access <br />
*Non-grid, local access to queue system<br />
<br />
| Both <br />
| €0.03 + VAT <br />
| €0.04 + VAT <br />
| TBD <br />
| <br />
| 23%<br />
|-<br />
! scope="row" | TUBITAK ULAKBIM <br />
| TR-10-ULAKBIM <br />
| Onur Temizsoylu<br>Feyza Eryol <br />
| Cloud + Batch <br />
| <br />
*Batch processing via Cream CE <br />
*VM Management using OpenNebula/OCCI <br />
*Infiniband connected Linux computing resources<br />
<br />
| Both <br />
| €0.15/hour 8 core 24 GB instance <br />
| TBC <br />
| TBD <br />
| <br />
| 18%<br />
|-<br />
! scope="row" | UIIP NASB <br />
| BY-UIIP <br />
| Serge Salamanka <br />
| Cloud + Batch <br />
| <br />
*Batch processing via UNICORE or QCG-Computing <br />
*Intel Xeon E5472 (3,0 GHz, 4 Cores) 50 nodes / 100 physical CPUs <br />
*400 virtual CPUs <br />
*400 Gb memory <br />
*8TB Disk storage – 9-12TB Storage servers <br />
*Infiniband DDR 20Gbit/sec]<br />
<br />
| Indirect <br />
| $0.077/hr 1 Core <br />
| $0.015/hr <br />
| $0.055/GB/month <br />
| $0.055/GB/month <br />
| 20%<br />
|-<br />
! scope="row" | Fraunhofer SCAI <br />
| SCAI <br />
| Horst Schwichtenberg <br />
| Cloud (Openstack) + Grid (gLite) <br />
| <br />
Cloud: <br />
<br />
*Intel Xeon X5570 @ 2.93GHz <br />
*Non-blocking QDR InfiniBand interconnect (40GBps)<br />
<br />
Grid: <br />
<br />
*AMD Opteron <br />
*DDR Infiniband<br />
<br />
| TBC <br />
| €0.02 <br />
| TBC <br />
| TBD <br />
| <br />
| 19%<br />
|-<br />
! scope="row" | CSC, IT Center for Science, Ltd. <br />
| CSC-cloud <br />
| Janne Ignatius, Luís Alves <br />
| Cloud (Openstack) <br />
| <br />
*Intel Xeon E5-2670 @ 2.6 GHz<br><br />
<br />
| Direct <br />
| N/A <br />
| TBC Soon <br />
| TBD <br />
| <br />
| 24%<br />
|}<br />
<br />
<br></div>Dmghttps://wiki.egi.eu/w/index.php?title=VT_EGI_Pay-for-Use_Technical_Details&diff=68230VT EGI Pay-for-Use Technical Details2014-06-19T08:55:10Z<p>Dmg: /* Resource Providers, Services and Prices */</p>
<hr />
<div>{{Template:EGI-PayForUse-PoC_menubar}} {{TOC right}} <br />
<br />
The following table provides an overview of the technical details of the resource and technology providers for the PoC. <br />
<br />
Further below are instructions for adding pricing into the accounting mechanisms. <br />
<br />
= Resource Providers, Services and Prices =<br />
<br />
{| class="wikitable"<br />
|-<br />
! scope="col" | NGI/RC <br />
! scope="col" | GOCDB Name <br />
! scope="col" | Contact <br />
! scope="col" | Service Type <br />
! scope="col" | Services and Resources <br />
! scope="col" | Preferred Payment Direct/Indirect <br />
! scope="col" | Grid Compute Price<br>[HEPSPEC06/hr] <br />
! scope="col" | Cloud Compute Price<br>[wallclock/hr] <br />
! scope="col" | Storage Price<br>[€/GB/Month] <br />
! scope="col" | VAT<br>[%]<br />
|-<br />
! scope="row" | Albert Einstein Center, Univ. of Bern <br />
| UNIBE-LHEP <br />
| Sigve Haug <br />
| Batch via CENTOS 5/6 ARC CE <br> <br />
| <br />
*500 cores (approx. 2 months in 2013) on highest priority; <br />
*Consumption preferably in one go, but distributed in time also possible <br />
*10-20 TB disk on local SE for consumption period via DPM SRM interface plus two weeks<br />
<br />
| Both <br />
| €0.03/hr <br />
| N/A <br />
| <br />
Free for PoC Other TBD <br />
<br />
| 8%<br />
|-<br />
! scope="row" | NGI_BG <br />
| BG01-IPP; BG03-NGCC <br />
| Todor Gurov<br>Emanouil Atanassov <br />
| <br />
Batch &amp; Grid: EMI-2/EMI-3 <br> Cloud: OpenStack - FedCloud <br />
<br />
| <br />
*16 NVIDIA Tesla M2090 6 GB + 48 core Intel Xeon E5679 @ 2.53GHz with 4 GB of RAM per core <br />
*576 cores Intel Xeon X5560 @ 2.8GHz with 1.5 GB of RAM per core <br />
*200 cores Intel Xeon E5430 @ 2.66GHz with 2 GB of RAM per core <br />
*MPI jobs <br />
*Non-blocking InfiniBand interconnect <br />
*140 TB of storage provisioned through Lustre FS (/home and /scratch systems), EMI dCache, EMI DPM, OpenStack Swift and OpenStack Cinder<br />
<br />
| Both <br />
| <br />
*€1.00/GPU/hour + VAT (GPU-enabled servers) <br />
*€0.07/core/hour + VAT (CPU only Intel Xeon X5560-based cluster) <br />
*€0.05/core/hour + VAT (CPU only Intel Xeon E5430-based cluster)<br />
<br />
| N/A <br />
| €0.04/GB/month (+VAT) via dCache. Maximal capacity 30TB.<br />
| 20%<br />
|-<br />
! scope="row" | CESGA-IBERGRID <br />
| CESGA <br />
| Alvaro Simon<br>Ivan Diaz <br />
| Batch + Cloud <br />
| <br />
*720 cores (Batch) AMD Opteron(tm) Processor 6174 2.2Ghz <br />
*240 cores (Cloud) Intel(R) Xeon(R) CPU E5520 2.27GHz <br />
*MPI jobs <br />
*RAM from 1GB to 32GB per core<br />
<br />
| Direct <br />
| <br />
€0.02/core/HEPSPEC06 hour <br />
<br />
| €0.04/core+2GB RAM/hour (+VAT) Example: <br />
*small: 1 core &amp; 2GB €0.04/hour <br />
*medium: 2 cores &amp; 4GB €0.08/hour <br />
*large: 4 cores &amp; 8GB €0.16/hour<br />
<br />
| €0.055/GB/month (+VAT) <br />
| 21%<br />
|-<br />
! scope="row" | IFCA-IBERGRID <br />
| IFCA-LCG2 <br />
| Jesus Marco <br />
| Batch (SLURM, SGE) + Cloud <br />
| <br />
*Batch via CREAM CE <br />
*Cloud via OpenStack <br />
*Up to 2500 cores in servers with 4, 8, 16, 40 physical cores <br />
*Infiniband access to other nodes and storage possible for MPI jobs <br />
*RAM from 4GB to 256GB per core (max 1TB)<br />
<br />
| Both, Direct preferred <br />
| <br />
*€0.05/core/hour for usual instances,&nbsp;&lt;0.01/core/HS06 hour (+ VAT) <br />
*€0.04/core/hour for multicore use (&gt;128 cores)<br />
<br />
| €0.03 <br />
| Minimal storage free, requests up to 100 TB high performance storage possible <br />
| 21%<br />
|-<br />
! scope="row" | Institute of Informatics of the Slovak Academy of Sciences (II SAS) <br />
| IISAS-FedCloud <br />
| <br />
Ladislav Hluchý<br> Viet Tran <br />
<br />
| Cloud (Openstack) <br />
| <br />
*120 cores (Cloud) Intel(R) Xeon(R) CPU E5570 2.93GHz <br />
*1, 2, 4, 8 cores VMs <br />
*RAM up to 4GB per core<br />
<br />
| Both <br />
| N/A <br />
| €0.06/core/h +VAT <br />
| TBD <br />
| 20%<br />
|-<br />
! scope="row" | Latvian Grid <br />
| IMCSUL <br />
| Kaspars Krampis <br />
| Cloud: OpenStack, GRID: ARC + SLURM <br />
| <br />
*CPU: Up to 256 cores, RAM up to 4GB per core<br />
<br />
| TBC <br />
| €0.07/core/hr + VAT <br />
| N/A <br />
| TBD <br />
| 21%<br />
|-<br />
! scope="row" | MASTER-UP Srl <br />
| UNI-PERUGIA <br />
| <br />
Alessandro Costantini<br>Antonio Lagana <br />
<br />
| EMI1/2, CEs, WNs, SE <br />
| <br />
*EMI1 middleware (phasing to EMI2) - CE's, WN's and SE. <br />
*10 nodes to start (intel 4core, ethernet connection) <br />
*Long experience in Molecular Science and Comp Chem, can provide support for such innovative applications and in the field of R&amp;D for combustion, energy and material science. <br />
*Managing a computer farm of 200 cores and 4TB of storage part of those resources are supporting EGI.<br />
<br />
| Direct <br />
| €0.05 <br />
| N/A <br />
| TBD <br />
| 22%<br />
|-<br />
! scope="row" | NGI_GRNET <br />
| All HG-* <br />
| Kostas Koumantaros <br />
| Cloud and Grid <br />
| <br />
*TBC<br />
<br />
| TBC <br />
| €0.05/hr + VAT <br />
| €0.05 + VAT <br />
| TBD <br />
| 23%<br />
|-<br />
! scope="row" | NGI-IT <br />
| <br />
INFN-BARI PRISMA-INFN-BARI (cloud) <br />
<br />
INFN-XX <br />
<br />
| Vincenzo Spinoso <br />
| EMI, OpenStack <br />
| <br />
*gLite MW stack: Cream CE, LFC, Storm SE, WMS+LB, BDII. <br />
*Cloud: OpenSTACK<br />
<br />
| TBD <br />
| €0.07/hr + VAT <br />
| €0.05 + VAT <br />
| TBD <br />
| 22%<br />
|-<br />
! scope="row" | 100 Percent IT <br />
| 100IT <br />
| David Blundell <br />
| Cloud (Openstack) <br />
| <br />
*Intel Xeon X5670 @ 2.93GHz <br />
*Cloud via OpenStack<br />
<br />
| Both <br />
| N/A <br />
| €0.07 + VAT <br />
| TBD <br />
| 20%<br />
|-<br />
! scope="row" | PL-Grid <br />
| CYFRONET-LCG2 <br />
| Mariusz Sterzel <br />
| Cloud + Batch via Cream / EMI <br />
| <br />
*Computing power, storage, and human support <br />
*Batch processing via Cream CE/EMI <br />
*Cloud access <br />
*Non-grid, local access to queue system<br />
<br />
| Both <br />
| €0.03 + VAT <br />
| €0.04 + VAT <br />
| TBD <br />
| 23%<br />
|-<br />
! scope="row" | TUBITAK ULAKBIM <br />
| TR-10-ULAKBIM <br />
| Onur Temizsoylu<br>Feyza Eryol <br />
| Cloud + Batch <br />
| <br />
*Batch processing via Cream CE <br />
*VM Management using OpenNebula/OCCI <br />
*Infiniband connected Linux computing resources<br />
<br />
| Both <br />
| €0.15/hour 8 core 24 GB instance <br />
| TBC <br />
| TBD <br />
| 18%<br />
|-<br />
! scope="row" | UIIP NASB <br />
| BY-UIIP <br />
| Serge Salamanka <br />
| Cloud + Batch <br />
| <br />
*Batch processing via UNICORE or Cream CE <br />
*Intel Xeon E5472 (3,0 GHz, 4 Cores) 50 nodes / 100 physical CPUs <br />
*400 virtual CPUs <br />
*400 Gb memory <br />
*8TB Disk storage – 9-12TB Storage servers <br />
*Infiniband DDR 20Gbit/sec]<br />
<br />
| Indirect <br />
| $0.077/hr (TBD) 1 Core <br />
| N/A<br />
| TBD<br />
| 20%<br />
|-<br />
! scope="row" | Fraunhofer SCAI <br />
| SCAI <br />
| Horst Schwichtenberg <br />
| Cloud (Openstack) + Grid (gLite) <br />
| <br />
Cloud: <br />
<br />
*Intel Xeon X5570 @ 2.93GHz <br />
*Non-blocking QDR InfiniBand interconnect (40GBps)<br />
<br />
Grid: <br />
<br />
*AMD Opteron <br />
*DDR Infiniband<br />
<br />
| TBC <br />
| €0.02 <br />
| TBC<br />
| TBD<br />
| 19%<br />
|}<br />
<br />
<br><br />
<br />
= Price in Accounting<br> =<br />
<br />
== Overview ==<br />
<br />
To test the mechanism for turning your accounting into a monetary charge charging rate needs to be defined. During the pilot this will be one number for your site in euros/HS06 hour. The number does not have to be real but obviously a ballpark figure would be good as it would give more realistic pilot reports. <br />
<br />
To define the charging rate the plan is to use a new feature of GOCDB which is known as ‘&lt;a href="https://wiki.egi.eu/wiki/GOCDB/Release4/Development/ExtensibilityMechanism"&gt;Key-Value Pair Property Bag Extensiblity Mechanism&lt;/a&gt;’. This allows the definition of arbitrary key-value pairs for a site or any service. If a community agrees a naming scheme for the keys and semantics for the values then they have a flexible and low overhead method of adding more information about anything defined in GOCDB. For anyone interested about more details this is documented &lt;a href="https://wiki.egi.eu/wiki/GOCDB/Release4/Development/ExtensibilityMechanism"&gt;here&lt;/a&gt;. <br />
<br />
== Instructions ==<br />
<br />
*If you do not have access rights to change features of your site please ask the site manager or a site admin to do it for you.<br />
<br />
#Go to GOCDB &lt;a href="https://goc.egi.eu"&gt;https://goc.egi.eu/&lt;/a&gt; <br />
#Navigate to your site. (My Sites) at the top of the left hand sidebar. <br />
#Just above the list of Services you should see a new box called 'Site Extension Properties' <br />
#At the bottom of this box there is a big green plus titled 'Add Properties' – click on it <br />
#This should show a window with two fields 'Property Name' and 'Property Value' <br />
#For 'Property Name' insert P4U_Pilot_Grid_CPU <br />
#For 'Property Value' insert a real number which will be interpreted as euros per HEPSPEC06 Hour. <br />
#Click the button 'Add Site Property' <br />
#If you go back to the site view you will see P4U_Pilot_Grid_CPU listed. You can edit or delete it.<br />
<br />
Once you have done this we will query GOCDB to pull the values for the pilot sites and the accounting portal will join them with usage data to produce a charging report.</div>Dmghttps://wiki.egi.eu/w/index.php?title=VT_EGI_Pay-for-Use_Technical_Details&diff=68229VT EGI Pay-for-Use Technical Details2014-06-19T08:52:48Z<p>Dmg: /* Resource Providers, Services and Prices */</p>
<hr />
<div>{{Template:EGI-PayForUse-PoC_menubar}} {{TOC right}} <br />
<br />
The following table provides an overview of the technical details of the resource and technology providers for the PoC. <br />
<br />
Further below are instructions for adding pricing into the accounting mechanisms. <br />
<br />
= Resource Providers, Services and Prices =<br />
<br />
{| class="wikitable"<br />
|-<br />
! scope="col" | NGI/RC <br />
! scope="col" | GOCDB Name <br />
! scope="col" | Contact <br />
! scope="col" | Service Type <br />
! scope="col" | Services and Resources <br />
! scope="col" | Preferred Payment Direct/Indirect <br />
! scope="col" | Grid Compute Price<br>[HEPSPEC06/hr] <br />
! scope="col" | Cloud Compute Price<br>[wallclock/hr] <br />
! scope="col" | Storage Price<br>[€/GB/Month] <br />
! scope="col" | VAT<br>[%]<br />
|-<br />
! scope="row" | Albert Einstein Center, Univ. of Bern <br />
| UNIBE-LHEP <br />
| Sigve Haug <br />
| Batch via CENTOS 5/6 ARC CE <br> <br />
| <br />
*500 cores (approx. 2 months in 2013) on highest priority; <br />
*Consumption preferably in one go, but distributed in time also possible <br />
*10-20 TB disk on local SE for consumption period via DPM SRM interface plus two weeks<br />
<br />
| Both <br />
| €0.03/hr <br />
| N/A <br />
| <br />
Free for PoC Other TBD <br />
<br />
| 8%<br />
|-<br />
! scope="row" | NGI_BG <br />
| BG01-IPP; BG03-NGCC <br />
| Todor Gurov<br>Emanouil Atanassov <br />
| <br />
Batch &amp; Grid: EMI-2/EMI-3 <br> Cloud: OpenStack - FedCloud <br />
<br />
| <br />
*16 NVIDIA Tesla M2090 6 GB + 48 core Intel Xeon E5679 @ 2.53GHz with 4 GB of RAM per core <br />
*576 cores Intel Xeon X5560 @ 2.8GHz with 1.5 GB of RAM per core <br />
*200 cores Intel Xeon E5430 @ 2.66GHz with 2 GB of RAM per core <br />
*MPI jobs <br />
*Non-blocking InfiniBand interconnect <br />
*140 TB of storage provisioned through Lustre FS (/home and /scratch systems), EMI dCache, EMI DPM, OpenStack Swift and OpenStack Cinder<br />
<br />
| Both <br />
| <br />
*€1.00/GPU/hour + VAT (GPU-enabled servers) <br />
*€0.07/core/hour + VAT (CPU only Intel Xeon X5560-based cluster) <br />
*€0.05/core/hour + VAT (CPU only Intel Xeon E5430-based cluster)<br />
<br />
| N/A <br />
| €0.04/GB/month (+VAT) via dCache<br />
| 20%<br />
|-<br />
! scope="row" | CESGA-IBERGRID <br />
| CESGA <br />
| Alvaro Simon<br>Ivan Diaz <br />
| Batch + Cloud <br />
| <br />
*720 cores (Batch) AMD Opteron(tm) Processor 6174 2.2Ghz <br />
*240 cores (Cloud) Intel(R) Xeon(R) CPU E5520 2.27GHz <br />
*MPI jobs <br />
*RAM from 1GB to 32GB per core<br />
<br />
| Direct <br />
| <br />
€0.02/core/HEPSPEC06 hour <br />
<br />
| €0.04/core+2GB RAM/hour (+VAT) Example: <br />
*small: 1 core &amp; 2GB €0.04/hour <br />
*medium: 2 cores &amp; 4GB €0.08/hour <br />
*large: 4 cores &amp; 8GB €0.16/hour<br />
<br />
| €0.055/GB/month (+VAT) <br />
| 21%<br />
|-<br />
! scope="row" | IFCA-IBERGRID <br />
| IFCA-LCG2 <br />
| Jesus Marco <br />
| Batch (SLURM, SGE) + Cloud <br />
| <br />
*Batch via CREAM CE <br />
*Cloud via OpenStack <br />
*Up to 2500 cores in servers with 4, 8, 16, 40 physical cores <br />
*Infiniband access to other nodes and storage possible for MPI jobs <br />
*RAM from 4GB to 256GB per core (max 1TB)<br />
<br />
| Both, Direct preferred <br />
| <br />
*€0.05/core/hour for usual instances,&nbsp;&lt;0.01/core/HS06 hour (+ VAT) <br />
*€0.04/core/hour for multicore use (&gt;128 cores)<br />
<br />
| €0.03 <br />
| Minimal storage free, requests up to 100 TB high performance storage possible <br />
| 21%<br />
|-<br />
! scope="row" | Institute of Informatics of the Slovak Academy of Sciences (II SAS) <br />
| IISAS-FedCloud <br />
| <br />
Ladislav Hluchý<br> Viet Tran <br />
<br />
| Cloud (Openstack) <br />
| <br />
*120 cores (Cloud) Intel(R) Xeon(R) CPU E5570 2.93GHz <br />
*1, 2, 4, 8 cores VMs <br />
*RAM up to 4GB per core<br />
<br />
| Both <br />
| N/A <br />
| €0.06/core/h +VAT <br />
| TBD <br />
| 20%<br />
|-<br />
! scope="row" | Latvian Grid <br />
| IMCSUL <br />
| Kaspars Krampis <br />
| Cloud: OpenStack, GRID: ARC + SLURM <br />
| <br />
*CPU: Up to 256 cores, RAM up to 4GB per core<br />
<br />
| TBC <br />
| €0.07/core/hr + VAT <br />
| N/A <br />
| TBD <br />
| 21%<br />
|-<br />
! scope="row" | MASTER-UP Srl <br />
| UNI-PERUGIA <br />
| <br />
Alessandro Costantini<br>Antonio Lagana <br />
<br />
| EMI1/2, CEs, WNs, SE <br />
| <br />
*EMI1 middleware (phasing to EMI2) - CE's, WN's and SE. <br />
*10 nodes to start (intel 4core, ethernet connection) <br />
*Long experience in Molecular Science and Comp Chem, can provide support for such innovative applications and in the field of R&amp;D for combustion, energy and material science. <br />
*Managing a computer farm of 200 cores and 4TB of storage part of those resources are supporting EGI.<br />
<br />
| Direct <br />
| €0.05 <br />
| N/A <br />
| TBD <br />
| 22%<br />
|-<br />
! scope="row" | NGI_GRNET <br />
| All HG-* <br />
| Kostas Koumantaros <br />
| Cloud and Grid <br />
| <br />
*TBC<br />
<br />
| TBC <br />
| €0.05/hr + VAT <br />
| €0.05 + VAT <br />
| TBD <br />
| 23%<br />
|-<br />
! scope="row" | NGI-IT <br />
| <br />
INFN-BARI PRISMA-INFN-BARI (cloud) <br />
<br />
INFN-XX <br />
<br />
| Vincenzo Spinoso <br />
| EMI, OpenStack <br />
| <br />
*gLite MW stack: Cream CE, LFC, Storm SE, WMS+LB, BDII. <br />
*Cloud: OpenSTACK<br />
<br />
| TBD <br />
| €0.07/hr + VAT <br />
| €0.05 + VAT <br />
| TBD <br />
| 22%<br />
|-<br />
! scope="row" | 100 Percent IT <br />
| 100IT <br />
| David Blundell <br />
| Cloud (Openstack) <br />
| <br />
*Intel Xeon X5670 @ 2.93GHz <br />
*Cloud via OpenStack<br />
<br />
| Both <br />
| N/A <br />
| €0.07 + VAT <br />
| TBD <br />
| 20%<br />
|-<br />
! scope="row" | PL-Grid <br />
| CYFRONET-LCG2 <br />
| Mariusz Sterzel <br />
| Cloud + Batch via Cream / EMI <br />
| <br />
*Computing power, storage, and human support <br />
*Batch processing via Cream CE/EMI <br />
*Cloud access <br />
*Non-grid, local access to queue system<br />
<br />
| Both <br />
| €0.03 + VAT <br />
| €0.04 + VAT <br />
| TBD <br />
| 23%<br />
|-<br />
! scope="row" | TUBITAK ULAKBIM <br />
| TR-10-ULAKBIM <br />
| Onur Temizsoylu<br>Feyza Eryol <br />
| Cloud + Batch <br />
| <br />
*Batch processing via Cream CE <br />
*VM Management using OpenNebula/OCCI <br />
*Infiniband connected Linux computing resources<br />
<br />
| Both <br />
| €0.15/hour 8 core 24 GB instance <br />
| TBC <br />
| TBD <br />
| 18%<br />
|-<br />
! scope="row" | UIIP NASB <br />
| BY-UIIP <br />
| Serge Salamanka <br />
| Cloud + Batch <br />
| <br />
*Batch processing via UNICORE or Cream CE <br />
*Intel Xeon E5472 (3,0 GHz, 4 Cores) 50 nodes / 100 physical CPUs <br />
*400 virtual CPUs <br />
*400 Gb memory <br />
*8TB Disk storage – 9-12TB Storage servers <br />
*Infiniband DDR 20Gbit/sec]<br />
<br />
| Indirect <br />
| $0.077/hr (TBD) 1 Core <br />
| N/A<br />
| TBD<br />
| 20%<br />
|-<br />
! scope="row" | Fraunhofer SCAI <br />
| SCAI <br />
| Horst Schwichtenberg <br />
| Cloud (Openstack) + Grid (gLite) <br />
| <br />
Cloud: <br />
<br />
*Intel Xeon X5570 @ 2.93GHz <br />
*Non-blocking QDR InfiniBand interconnect (40GBps)<br />
<br />
Grid: <br />
<br />
*AMD Opteron <br />
*DDR Infiniband<br />
<br />
| TBC <br />
| €0.02 <br />
| TBC<br />
| TBD<br />
| 19%<br />
|}<br />
<br />
<br><br />
<br />
= Price in Accounting<br> =<br />
<br />
== Overview ==<br />
<br />
To test the mechanism for turning your accounting into a monetary charge charging rate needs to be defined. During the pilot this will be one number for your site in euros/HS06 hour. The number does not have to be real but obviously a ballpark figure would be good as it would give more realistic pilot reports. <br />
<br />
To define the charging rate the plan is to use a new feature of GOCDB which is known as ‘&lt;a href="https://wiki.egi.eu/wiki/GOCDB/Release4/Development/ExtensibilityMechanism"&gt;Key-Value Pair Property Bag Extensiblity Mechanism&lt;/a&gt;’. This allows the definition of arbitrary key-value pairs for a site or any service. If a community agrees a naming scheme for the keys and semantics for the values then they have a flexible and low overhead method of adding more information about anything defined in GOCDB. For anyone interested about more details this is documented &lt;a href="https://wiki.egi.eu/wiki/GOCDB/Release4/Development/ExtensibilityMechanism"&gt;here&lt;/a&gt;. <br />
<br />
== Instructions ==<br />
<br />
*If you do not have access rights to change features of your site please ask the site manager or a site admin to do it for you.<br />
<br />
#Go to GOCDB &lt;a href="https://goc.egi.eu"&gt;https://goc.egi.eu/&lt;/a&gt; <br />
#Navigate to your site. (My Sites) at the top of the left hand sidebar. <br />
#Just above the list of Services you should see a new box called 'Site Extension Properties' <br />
#At the bottom of this box there is a big green plus titled 'Add Properties' – click on it <br />
#This should show a window with two fields 'Property Name' and 'Property Value' <br />
#For 'Property Name' insert P4U_Pilot_Grid_CPU <br />
#For 'Property Value' insert a real number which will be interpreted as euros per HEPSPEC06 Hour. <br />
#Click the button 'Add Site Property' <br />
#If you go back to the site view you will see P4U_Pilot_Grid_CPU listed. You can edit or delete it.<br />
<br />
Once you have done this we will query GOCDB to pull the values for the pilot sites and the accounting portal will join them with usage data to produce a charging report.</div>Dmghttps://wiki.egi.eu/w/index.php?title=VT_EGI_Pay-for-Use_Technical_Details&diff=65208VT EGI Pay-for-Use Technical Details2014-02-28T09:33:59Z<p>Dmg: /* Resource Providers, Services and Prices */</p>
<hr />
<div>{{Template:EGI-PayForUse-PoC_menubar}} {{TOC right}} <br />
<br />
The following table provides an overview of the technical details of the resource and technology providers for the PoC. <br />
<br />
Further below are instructions for adding pricing into the accounting mechanisms. <br />
<br />
= Resource Providers, Services and Prices =<br />
<br />
{| class="wikitable"<br />
|-<br />
! scope="col" | NGI/RC <br />
! scope="col" | GOCDB Name <br />
! scope="col" | Contact <br />
! scope="col" | Service Type <br />
! scope="col" | Services and Resources <br />
! scope="col" | Preferred Payment Direct/Indirect <br />
! scope="col" | Grid Compute Price<br>[HEPSPEC06/hr] <br />
! scope="col" | Cloud Compute Price<br>[TBD] <br />
! scope="col" | Storage Price<br>[€/GB]<br />
|-<br />
! scope="row" | Albert Einstein Center, Univ. of Bern <br />
| UNIBE-LHEP <br />
| Sigve Haug <br />
| Batch via CENTOS 5/6 ARC CE <br> <br />
| <br />
*500 cores (approx. 2 months in 2013) on highest priority; <br />
*Consumption preferably in one go, but distributed in time also possible <br />
*10-20 TB disk on local SE for consumption period via DPM SRM interface plus two weeks<br />
<br />
| Both <br />
| €0.03/hr <br />
| <br />
| <br />
*Free for PoC <br />
*Other TBD<br />
<br />
|-<br />
! scope="row" | NGI_BG <br />
| BG01-IPP; BG03-NGCC <br />
| Todor Gurov<br>Emanouil Atanassov <br />
| <br />
Batch &amp; Grid: EMI-2/EMI-3 <br> Cloud: OpenStack - FedCloud <br />
<br />
| <br />
*16 NVIDIA Tesla M2090 6 GB + 48 core Intel Xeon E5679 @ 2.53GHz with 4 GB of RAM per core <br />
*576 cores Intel Xeon X5560 @ 2.8GHz with 1.5 GB of RAM per core <br />
*200 cores Intel Xeon E5430 @ 2.66GHz with 2 GB of RAM per core <br />
*MPI jobs <br />
*Non-blocking InfiniBand interconnect <br />
*140 TB of storage provisioned through Lustre FS (/home and /scratch systems), EMI dCache, EMI DPM, OpenStack Swift and OpenStack Cinder<br />
<br />
| Both <br />
| <br />
*€1.00/GPU/hour + 20% VAT (GPU-enabled servers) <br />
*€0.07/core/hour + 20% VAT (CPU only Intel Xeon X5560-based cluster) <br />
*€0.05/core/hour + 20% VAT (CPU only Intel Xeon E5430-based cluster)<br />
<br />
| <br />
| <br />
|-<br />
! scope="row" | CESGA-IBERGRID <br />
| CESGA <br />
| Alvaro Simon<br>Ivan Diaz <br />
| Batch + Cloud <br />
| <br />
*720 cores (Batch) AMD Opteron(tm) Processor 6174 2.2Ghz <br />
*240 cores (Cloud) Intel(R) Xeon(R) CPU E5520 2.27GHz <br />
*MPI jobs <br />
*RAM from 4GB to 32GB per core<br />
<br />
| Direct <br />
| <br />
€0.02/core/HEPSPEC06 hour <br />
<br />
| <br />
| <br />
|-<br />
! scope="row" | IFCA-IBERGRID <br />
| IFCA-LCG2 <br />
| Jesus Marco <br />
| Batch (SLURM, SGE) + Cloud <br />
| <br />
*Batch via CREAM CE <br />
*Cloud via OpenStack <br />
*Up to 2500 cores in servers with 4, 8, 16, 40 physical cores <br />
*Infiniband access to other nodes and storage possible for MPI jobs <br />
*RAM from 4GB to 256GB per core (max 1TB)<br />
<br />
| Both, Direct preferred <br />
| <br />
*€0.05/core/hour for usual instances,&nbsp;&lt;0.01/core/HS06 hour (+ VA <br />
*€0.04/core/hour for multicore use (&gt;128 cores)<br />
<br />
| <br />
| Minimal storage free, requests up to 100 TB high performance storage possible<br />
|-<br />
! scope="row" | Institute of Informatics of the Slovak Academy of Sciences (II SAS) <br />
| IISAS-FedCloud <br />
| <br />
Ladislav Hluchý<br> Viet Tran <br />
<br />
| Cloud <br />
| <br />
*120 cores (Cloud) Intel(R) Xeon(R) CPU E5570 2.93GHz <br />
*1, 2, 4, 8 cores VMs <br />
*RAM up to 4GB per core<br />
<br />
| Both <br />
| €0.06/core/hr <br />
| <br />
| <br />
|-<br />
! scope="row" | Latvian Grid <br />
| IMCSUL <br />
| Kaspars Krampis <br />
| Cloud: OpenStack, GRID: ARC + SLURM <br />
| <br />
*CPU: Up to 256 cores, RAM up to 4GB per core<br />
<br />
| TBC <br />
| ~€0.07/core/hr + 21% VAT (TBC)<br />
| <br />
| <br />
|-<br />
! scope="row" | MASTER-UP Srl <br />
| TBC <br />
| <br />
Alessandro Costantini<br>Antonio Lagana <br />
<br />
| EMI1/2, CEs, WNs, SE <br />
| <br />
*EMI1 middleware (phasing to EMI2) - CE's, WN's and SE. <br />
*10 nodes to start (intel 4core, ethernet connection) <br />
*Long experience in Molecular Science and Comp Chem, can provide support for such innovative applications and in the field of R&amp;D for combustion, energy and material science. <br />
*Managing a computer farm of 200 cores and 4TB of storage part of those resources are supporting EGI.<br />
<br />
| Direct <br />
| To be evaluated <br />
| <br />
| <br />
|-<br />
! scope="row" | NGI_GRNET <br />
| All HG-* <br />
| Kostas Koumantaros <br />
| Cloud and Grid <br />
| <br />
*TBC<br />
<br />
| TBC <br />
| €0.05/hr + 23% VAT <br />
| <br />
| <br />
|-<br />
! scope="row" | NGI-IT <br />
| <br />
INFN-BARI PRISMA-INFN-BARI (cloud) <br />
<br />
INFN-XX <br />
<br />
| Vincenzo Spinoso <br />
| EMI, OpenStack <br />
| <br />
*gLite MW stack: Cream CE, LFC, Storm SE, WMS+LB, BDII. <br />
*Cloud: OpenSTACK<br />
<br />
| TBD <br />
| €0.07/hr + 22% VAT <br />
| TBD <br />
| <br />
|-<br />
! scope="row" | 100 Percent IT <br />
| 100IT <br />
| David Blundell <br />
| Cloud (Openstack)<br />
|<br />
*Intel Xeon X5670 @ 2.93GHz<br />
*Cloud via OpenStack<br />
| Both <br />
| <br />
| €0.07 + 20% VAT <br />
|<br />
|-<br />
! scope="row" | PL-Grid <br />
| TBC <br />
| Mariusz Sterzel <br />
| Cloud + Batch via Cream / EMI <br />
| <br />
*Computing power, storage, and human support <br />
*Batch processing via Cream CE/EMI <br />
*Cloud access <br />
*Non-grid, local access to queue system<br />
<br />
| Both <br />
| Capacity and price are a subject to negotiation <br />
| <br />
| <br />
|-<br />
! scope="row" | TUBITAK ULAKBIM <br />
| TR-10-ULAKBIM <br />
| Onur Temizsoylu<br>Feyza Eryol <br />
| Cloud + Batch <br />
| <br />
*Batch processing via Cream CE <br />
*VM Management using OpenNebula/OCCI <br />
*Infiniband connected Linux computing resources<br />
<br />
| Both <br />
| €0.15/hour 8 core 24 GB instance <br />
| <br />
| <br />
|-<br />
! scope="row" | UIIP NASB <br />
| BY-UIIP <br />
| Serge Salamanka <br />
| Cloud + Batch <br />
| <br />
*Batch processing via UNICORE or Cream CE <br />
*Intel Xeon E5472 (3,0 GHz, 4 Cores) 50 nodes / 100 physical CPUs <br />
*400 virtual CPUs <br />
*400 Gb memory <br />
*8TB Disk storage – 9-12TB Storage servers <br />
*Infiniband DDR 20Gbit/sec]<br />
<br />
| Indirect <br />
| $0.077/hr (TBD) 1 Core <br />
| <br />
| <br />
|-<br />
! scope="row" | Fraunhofer SCAI <br />
| SCAI <br />
| Horst Schwichtenberg <br />
| Cloud (Openstack) + Batch <br />
| <br />
*Intel Xeon X5570 @ 2.93GHz <br />
*Non-blocking QDR InfiniBand interconnect (40GBps)<br />
<br />
| TBC <br />
| TBC <br />
| <br />
| <br />
|}<br />
<br />
<br><br />
<br />
= Price in Accounting<br> =<br />
<br />
== Overview ==<br />
<br />
To test the mechanism for turning your accounting into a monetary charge charging rate needs to be defined. During the pilot this will be one number for your site in euros/HS06 hour. The number does not have to be real but obviously a ballpark figure would be good as it would give more realistic pilot reports. <br />
<br />
To define the charging rate the plan is to use a new feature of GOCDB which is known as ‘[https://wiki.egi.eu/wiki/GOCDB/Release4/Development/ExtensibilityMechanism Key-Value Pair Property Bag Extensiblity Mechanism]’. This allows the definition of arbitrary key-value pairs for a site or any service. If a community agrees a naming scheme for the keys and semantics for the values then they have a flexible and low overhead method of adding more information about anything defined in GOCDB. For anyone interested about more details this is documented [https://wiki.egi.eu/wiki/GOCDB/Release4/Development/ExtensibilityMechanism here]. <br />
<br />
== Instructions ==<br />
<br />
*If you do not have access rights to change features of your site please ask the site manager or a site admin to do it for you.<br />
<br />
#Go to the test GOCDB [https://gocdb-test.esc.rl.ac.uk/v5 https://gocdb-test.esc.rl.ac.uk/v5] which contains a copy of the production database. The same access rights apply but you will not change anything in the production database. <br />
#Navigate to your site. (My Sites) at the top of the left hand sidebar. <br />
#Just above the list of Services you should see a new box called 'Site Extension Properties' <br />
#At the bottom of this box there is a big green plus titled 'Add Properties' – click on it <br />
#This should show a window with two fields 'Property Name' and 'Property Value' <br />
#For 'Property Name' insert TEST_CHARGE <br />
#For 'Property Value' insert a real number which will be interpreted as euros per HEPSPEC06 Hour. <br />
#Click the button 'Add Site Property' <br />
#If you go back to the site view you will see TEST_CHARGE listed. You can edit or delete it.<br />
<br />
Once you have done this we will query GOCDB to pull the values for the pilot sites and the accounting portal will join them with usage data to produce a charging report.</div>Dmghttps://wiki.egi.eu/w/index.php?title=VT_EGI_Pay-for-Use_Technical_Details&diff=64534VT EGI Pay-for-Use Technical Details2014-02-13T11:12:20Z<p>Dmg: </p>
<hr />
<div>{{Template:EGI-PayForUse-PoC_menubar}} {{TOC right}} <br />
<br />
The following table provides an overview of the technical details of the resource and technology providers for the PoC. <br />
<br />
Further below are instructions for adding pricing into the accounting mechanisms. <br />
<br />
[1] All prices provided are interpreted as Euro per HEPSPEC06 hour. <br />
<br />
= Resource Providers, Services and Prices =<br />
<br />
{| class="wikitable"<br />
|-<br />
! scope="col" | NGI/RC <br />
! scope="col" | GOCDB Name <br />
! scope="col" | Contact <br />
! scope="col" | Service Type <br />
! scope="col" | Services and Resources <br />
! scope="col" | Preferred Payment Direct/Indirect <br />
! scope="col" | Price [1]<br />
|-<br />
! scope="row" | Albert Einstein Center, Univ. of Bern <br />
| UNIBE-LHEP <br />
| <br />
| Batch via CENTOS 5/6 ARC CE <br> <br />
| <br />
*500 cores (approx. 2 months in 2013) on highest priority; <br />
*Consumption preferably in one go, but distributed in time also possible <br />
*10-20 TB disk on local SE for consumption period via DPM SRM interface plus two weeks<br />
<br />
| Both <br />
| €0.03/hr; Storage free<br />
|-<br />
! scope="row" | NGI_BG<br />
|<br />
BG01-IPP<br />
BG08-MADARA<br />
| <br />
|<br />
Batch & Grid: EMI-2/EMI-3 <br /><br />
Cloud: OpenStack - FedCloud<br />
|<br />
* 16 NVIDIA Tesla M2090 6 GB + 48 core Intel Xeon E5679 @ 2.53GHz with 4 GB of RAM per core<br />
* 576 cores Intel Xeon X5560 @ 2.8GHz with 1.5 GB of RAM per core<br />
* 800 cores Intel Xeon E5520 @ 2.26GHz with 1 GB of RAM per core<br />
* MPI jobs<br />
* Non-blocking InfiniBand interconnect<br />
* 240 TB of storage provisioned through Lustre FS (/home and /scratch systems), EMI dCache, EMI DPM, OpenStack Swift and OpenStack Cinder<br />
| Both<br />
|<br />
€1.00/GPU/hour + 20% VAT (GPU-enabled servers) <br /><br />
€0.07/core/hour + 20% VAT (CPU only Intel Xeon X5560-based cluster) <br /><br />
€0.06/core/hour + 20% VAT (CPU only Intel Xeon E5520-based cluster)<br />
|-<br />
! scope="row" | CESGA-IBERGRID <br />
| CESGA <br />
| <br />
| Batch + Cloud <br />
| <br />
*720 cores (Batch) AMD Opteron(tm) Processor 6174 2.2Ghz <br />
*240 cores (Cloud) Intel(R) Xeon(R) CPU E5520 2.27GHz <br />
*MPI jobs <br />
*RAM from 4GB to 32GB per core<br />
<br />
| Direct <br />
| <br />
€0.02/core/HEPSPEC06 hour <br />
<br />
|-<br />
! scope="row" | IFCA-IBERGRID <br />
| TBD <br />
| <br />
| Batch (SLURM, SGE) or Cloud (FedCloud soon) <br />
| <br />
*Up to 2500 cores in servers with 4, 8, 16, 40 physical cores <br />
*Infiniband access to other nodes and storage possible for MPI jobs <br />
*RAM from 4GB to 256GB per core (max 1TB)<br />
<br />
| Direct <br />
| <br />
€0.05/core/hour for usual instances <br />
<br />
<span style="line-height: 1.5em;">€0.04/core/hour for multicore use (&gt;128 cores)</span> <br />
<br />
<span style="line-height: 1.5em;">Minimal storage free, requests up to 100 TB high performance storage possible</span> <br />
<br />
|-<br />
! scope="row" | Institute of Informatics of the Slovak Academy of Sciences <br />
| IISAS-FedCloud <br />
| <br />
| Cloud <br />
| <br />
*TBC<br />
<br />
| TBC <br />
| TBC<br />
|-<br />
! scope="row" | Latvian Grid <br />
| IMCSUL <br />
| <br />
| TBC <br />
| <br />
*TBC<br />
<br />
| TBC <br />
| TBC<br />
|-<br />
! scope="row" | Leibniz SC Centre <br />
| LRZ <br />
| <br />
| Globus <br />
| <br />
*Linux Cluster through Globus: GRAM5, GridFTP, Gsissh<br />
<br />
| Indirect <br />
| Price and capacity to be detailed later<br />
|-<br />
! scope="row" | MASTER-UP Srl <br />
| TBC <br />
| <br />
| EMI1/2, CEs, WNs, SE <br />
| <br />
*EMI1 middleware (phasing to EMI2) - CE's, WN's and SE. <br />
*10 nodes to start (intel 4core, ethernet connection) <br />
*Long experience in Molecular Science and Comp Chem, can provide support for such innovative applications and in the field of R&amp;D for combustion, energy and material science. <br />
*Managing a computer farm of 200 cores and 4TB of storage part of those resources are supporting EGI.<br />
<br />
| Direct <br />
| To be evaluated<br />
|-<br />
! scope="row" | NGI-DE <br />
| TBC <br />
| <br />
| UNICORE <br />
| <br />
*Cluster via UNICORE <br />
*Others can be offered (TBD)<br />
<br />
| Both <br />
| €0.06/hr + 19% VAT<br />
|-<br />
! scope="row" | NGI_GRNET <br />
| All HG-* <br />
| <br />
| Cloud and Grid <br />
| <br />
*TBC<br />
<br />
| TBC <br />
| €0.05/hr + 23% VAT<br />
|-<br />
! scope="row" | NGI-IT <br />
| <br />
INFN-BARI <br />
<br />
PRISMA-INFN-BARI (cloud) <br />
<br />
INFN-XX <br />
<br />
| <br />
| EMI, OpenStack <br />
| <br />
*gLite MW stack: Cream CE, LFC, Storm SE, WMS+LB, BDII. <br />
*Cloud: OpenSTACK<br />
<br />
| TBD <br />
| <br />
€0.07/hr + 22% VAT(grid) <br />
<br />
TDB (cloud) <br />
<br />
|-<br />
! scope="row" | PL-Grid <br />
| TBC <br />
| <br />
| Cloud + Batch via Cream / EMI <br />
| <br />
*Computing power, storage, and human support <br />
*Batch processing via Cream CE/EMI <br />
*Cloud access <br />
*Non-grid, local access to queue system<br />
<br />
| Both <br />
| Capacity and price are a subject to negotiation<br />
|-<br />
! scope="row" | TUBITAK ULAKBIM <br />
| TR-10-ULAKBIM <br />
| <br />
| Cloud + Batch <br />
| <br />
*Batch processing via Cream CE <br />
*VM Management using OpenNebula/OCCI <br />
*Infiniband connected Linux computing resources<br />
<br />
| Both <br />
| €0.15/hour 8 core 24 GB instance<br />
|-<br />
! scope="row" | UIIP NASB <br />
| BY-UIIP <br />
| <br />
| Cloud + Batch <br />
| <br />
*Batch processing via UNICORE or Cream CE <br />
*Intel Xeon E5472 (3,0 GHz, 4 Cores) 50 nodes / 100 physical CPUs <br />
*400 virtual CPUs <br />
*400 Gb memory <br />
*8TB Disk storage – 9-12TB Storage servers <br />
*Infiniband DDR 20Gbit/sec]<br />
<br />
| Indirect <br />
| $0.077/hr (TBD) 1 Core<br />
|}<br />
<br />
<br><br />
<br />
= Price in Accounting<br> =<br />
<br />
== Overview ==<br />
<br />
To test the mechanism for turning your accounting into a monetary charge charging rate needs to be defined. During the pilot this will be one number for your site in euros/HS06 hour. The number does not have to be real but obviously a ballpark figure would be good as it would give more realistic pilot reports. <br />
<br />
To define the charging rate the plan is to use a new feature of GOCDB which is known as ‘[https://wiki.egi.eu/wiki/GOCDB/Release4/Development/ExtensibilityMechanism Key-Value Pair Property Bag Extensiblity Mechanism]’. This allows the definition of arbitrary key-value pairs for a site or any service. If a community agrees a naming scheme for the keys and semantics for the values then they have a flexible and low overhead method of adding more information about anything defined in GOCDB. For anyone interested about more details this is documented [https://wiki.egi.eu/wiki/GOCDB/Release4/Development/ExtensibilityMechanism here]. <br />
<br />
== Instructions ==<br />
<br />
*If you do not have access rights to change features of your site please ask the site manager or a site admin to do it for you.<br />
<br />
#Go to the test GOCDB [https://gocdb-test.esc.rl.ac.uk/v5 https://gocdb-test.esc.rl.ac.uk/v5] which contains a copy of the production database. The same access rights apply but you will not change anything in the production database. <br />
#Navigate to your site. (My Sites) at the top of the left hand sidebar. <br />
#Just above the list of Services you should see a new box called 'Site Extension Properties' <br />
#At the bottom of this box there is a big green plus titled 'Add Properties' – click on it <br />
#This should show a window with two fields 'Property Name' and 'Property Value' <br />
#For 'Property Name' insert TEST_CHARGE <br />
#For 'Property Value' insert a real number which will be interpreted as euros per HEPSPEC06 Hour. <br />
#Click the button 'Add Site Property' <br />
#If you go back to the site view you will see TEST_CHARGE listed. You can edit or delete it.<br />
<br />
Once you have done this we will query GOCDB to pull the values for the pilot sites and the accounting portal will join them with usage data to produce a charging report.</div>Dmghttps://wiki.egi.eu/w/index.php?title=VT_EGI_Pay-for-Use_Technical_Details&diff=64533VT EGI Pay-for-Use Technical Details2014-02-13T11:11:20Z<p>Dmg: </p>
<hr />
<div>{{Template:EGI-PayForUse-PoC_menubar}} {{TOC right}} <br />
<br />
The following table provides an overview of the technical details of the resource and technology providers for the PoC. <br />
<br />
Further below are instructions for adding pricing into the accounting mechanisms. <br />
<br />
[1] All prices provided are interpreted as Euro per HEPSPEC06 hour. <br />
<br />
= Resource Providers, Services and Prices =<br />
<br />
{| class="wikitable"<br />
|-<br />
! scope="col" | NGI/RC <br />
! scope="col" | GOCDB Name <br />
! scope="col" | Contact <br />
! scope="col" | Service Type <br />
! scope="col" | Services and Resources <br />
! scope="col" | Preferred Payment Direct/Indirect <br />
! scope="col" | Price [1]<br />
|-<br />
! scope="row" | Albert Einstein Center, Univ. of Bern <br />
| UNIBE-LHEP <br />
| <br />
| Batch via CENTOS 5/6 ARC CE <br> <br />
| <br />
*500 cores (approx. 2 months in 2013) on highest priority; <br />
*Consumption preferably in one go, but distributed in time also possible <br />
*10-20 TB disk on local SE for consumption period via DPM SRM interface plus two weeks<br />
<br />
| Both <br />
| €0.03/hr; Storage free<br />
|-<br />
! scope="row" | NGI_BG<br />
|<br />
BG01-IPP<br />
BG08-MADARA<br />
| <br />
|<br />
Batch & Grid: EMI-2/EMI-3 <br /><br />
Cloud: OpenStack - FedCloud<br />
|<br />
* 16 NVIDIA Tesla M2090 6 GB + 48 core Intel Xeon E5679 @ 2.53GHz with 4 GB of RAM per core<br />
* 576 cores Intel Xeon X5560 @ 2.8GHz with 1.5 GB of RAM per core<br />
* 800 cores Intel Xeon E5520 @ 2.26GHz with 1 GB of RAM per core<br />
* MPI jobs<br />
* Non-blocking InfiniBand interconnect<br />
* 240 TB of storage provisioned through EMI dCache, EMI DPM and Lustre FS /home and /scratch systems, OpenStack Swift and OpenStack Cinder<br />
| Both<br />
|<br />
€1.00/GPU/hour + 20% VAT (GPU-enabled servers) <br /><br />
€0.07/core/hour + 20% VAT (CPU only Intel Xeon X5560-based cluster) <br /><br />
€0.06/core/hour + 20% VAT (CPU only Intel Xeon E5520-based cluster)<br />
|-<br />
! scope="row" | CESGA-IBERGRID <br />
| CESGA <br />
| <br />
| Batch + Cloud <br />
| <br />
*720 cores (Batch) AMD Opteron(tm) Processor 6174 2.2Ghz <br />
*240 cores (Cloud) Intel(R) Xeon(R) CPU E5520 2.27GHz <br />
*MPI jobs <br />
*RAM from 4GB to 32GB per core<br />
<br />
| Direct <br />
| <br />
€0.02/core/HEPSPEC06 hour <br />
<br />
|-<br />
! scope="row" | IFCA-IBERGRID <br />
| TBD <br />
| <br />
| Batch (SLURM, SGE) or Cloud (FedCloud soon) <br />
| <br />
*Up to 2500 cores in servers with 4, 8, 16, 40 physical cores <br />
*Infiniband access to other nodes and storage possible for MPI jobs <br />
*RAM from 4GB to 256GB per core (max 1TB)<br />
<br />
| Direct <br />
| <br />
€0.05/core/hour for usual instances <br />
<br />
<span style="line-height: 1.5em;">€0.04/core/hour for multicore use (&gt;128 cores)</span> <br />
<br />
<span style="line-height: 1.5em;">Minimal storage free, requests up to 100 TB high performance storage possible</span> <br />
<br />
|-<br />
! scope="row" | Institute of Informatics of the Slovak Academy of Sciences <br />
| IISAS-FedCloud <br />
| <br />
| Cloud <br />
| <br />
*TBC<br />
<br />
| TBC <br />
| TBC<br />
|-<br />
! scope="row" | Latvian Grid <br />
| IMCSUL <br />
| <br />
| TBC <br />
| <br />
*TBC<br />
<br />
| TBC <br />
| TBC<br />
|-<br />
! scope="row" | Leibniz SC Centre <br />
| LRZ <br />
| <br />
| Globus <br />
| <br />
*Linux Cluster through Globus: GRAM5, GridFTP, Gsissh<br />
<br />
| Indirect <br />
| Price and capacity to be detailed later<br />
|-<br />
! scope="row" | MASTER-UP Srl <br />
| TBC <br />
| <br />
| EMI1/2, CEs, WNs, SE <br />
| <br />
*EMI1 middleware (phasing to EMI2) - CE's, WN's and SE. <br />
*10 nodes to start (intel 4core, ethernet connection) <br />
*Long experience in Molecular Science and Comp Chem, can provide support for such innovative applications and in the field of R&amp;D for combustion, energy and material science. <br />
*Managing a computer farm of 200 cores and 4TB of storage part of those resources are supporting EGI.<br />
<br />
| Direct <br />
| To be evaluated<br />
|-<br />
! scope="row" | NGI-DE <br />
| TBC <br />
| <br />
| UNICORE <br />
| <br />
*Cluster via UNICORE <br />
*Others can be offered (TBD)<br />
<br />
| Both <br />
| €0.06/hr + 19% VAT<br />
|-<br />
! scope="row" | NGI_GRNET <br />
| All HG-* <br />
| <br />
| Cloud and Grid <br />
| <br />
*TBC<br />
<br />
| TBC <br />
| €0.05/hr + 23% VAT<br />
|-<br />
! scope="row" | NGI-IT <br />
| <br />
INFN-BARI <br />
<br />
PRISMA-INFN-BARI (cloud) <br />
<br />
INFN-XX <br />
<br />
| <br />
| EMI, OpenStack <br />
| <br />
*gLite MW stack: Cream CE, LFC, Storm SE, WMS+LB, BDII. <br />
*Cloud: OpenSTACK<br />
<br />
| TBD <br />
| <br />
€0.07/hr + 22% VAT(grid) <br />
<br />
TDB (cloud) <br />
<br />
|-<br />
! scope="row" | PL-Grid <br />
| TBC <br />
| <br />
| Cloud + Batch via Cream / EMI <br />
| <br />
*Computing power, storage, and human support <br />
*Batch processing via Cream CE/EMI <br />
*Cloud access <br />
*Non-grid, local access to queue system<br />
<br />
| Both <br />
| Capacity and price are a subject to negotiation<br />
|-<br />
! scope="row" | TUBITAK ULAKBIM <br />
| TR-10-ULAKBIM <br />
| <br />
| Cloud + Batch <br />
| <br />
*Batch processing via Cream CE <br />
*VM Management using OpenNebula/OCCI <br />
*Infiniband connected Linux computing resources<br />
<br />
| Both <br />
| €0.15/hour 8 core 24 GB instance<br />
|-<br />
! scope="row" | UIIP NASB <br />
| BY-UIIP <br />
| <br />
| Cloud + Batch <br />
| <br />
*Batch processing via UNICORE or Cream CE <br />
*Intel Xeon E5472 (3,0 GHz, 4 Cores) 50 nodes / 100 physical CPUs <br />
*400 virtual CPUs <br />
*400 Gb memory <br />
*8TB Disk storage – 9-12TB Storage servers <br />
*Infiniband DDR 20Gbit/sec]<br />
<br />
| Indirect <br />
| $0.077/hr (TBD) 1 Core<br />
|}<br />
<br />
<br><br />
<br />
= Price in Accounting<br> =<br />
<br />
== Overview ==<br />
<br />
To test the mechanism for turning your accounting into a monetary charge charging rate needs to be defined. During the pilot this will be one number for your site in euros/HS06 hour. The number does not have to be real but obviously a ballpark figure would be good as it would give more realistic pilot reports. <br />
<br />
To define the charging rate the plan is to use a new feature of GOCDB which is known as ‘[https://wiki.egi.eu/wiki/GOCDB/Release4/Development/ExtensibilityMechanism Key-Value Pair Property Bag Extensiblity Mechanism]’. This allows the definition of arbitrary key-value pairs for a site or any service. If a community agrees a naming scheme for the keys and semantics for the values then they have a flexible and low overhead method of adding more information about anything defined in GOCDB. For anyone interested about more details this is documented [https://wiki.egi.eu/wiki/GOCDB/Release4/Development/ExtensibilityMechanism here]. <br />
<br />
== Instructions ==<br />
<br />
*If you do not have access rights to change features of your site please ask the site manager or a site admin to do it for you.<br />
<br />
#Go to the test GOCDB [https://gocdb-test.esc.rl.ac.uk/v5 https://gocdb-test.esc.rl.ac.uk/v5] which contains a copy of the production database. The same access rights apply but you will not change anything in the production database. <br />
#Navigate to your site. (My Sites) at the top of the left hand sidebar. <br />
#Just above the list of Services you should see a new box called 'Site Extension Properties' <br />
#At the bottom of this box there is a big green plus titled 'Add Properties' – click on it <br />
#This should show a window with two fields 'Property Name' and 'Property Value' <br />
#For 'Property Name' insert TEST_CHARGE <br />
#For 'Property Value' insert a real number which will be interpreted as euros per HEPSPEC06 Hour. <br />
#Click the button 'Add Site Property' <br />
#If you go back to the site view you will see TEST_CHARGE listed. You can edit or delete it.<br />
<br />
Once you have done this we will query GOCDB to pull the values for the pilot sites and the accounting portal will join them with usage data to produce a charging report.</div>Dmghttps://wiki.egi.eu/w/index.php?title=VT_EGI_Pay-for-Use_Technical_Details&diff=64532VT EGI Pay-for-Use Technical Details2014-02-13T11:10:24Z<p>Dmg: </p>
<hr />
<div>{{Template:EGI-PayForUse-PoC_menubar}} {{TOC right}} <br />
<br />
The following table provides an overview of the technical details of the resource and technology providers for the PoC. <br />
<br />
Further below are instructions for adding pricing into the accounting mechanisms. <br />
<br />
[1] All prices provided are interpreted as Euro per HEPSPEC06 hour. <br />
<br />
= Resource Providers, Services and Prices =<br />
<br />
{| class="wikitable"<br />
|-<br />
! scope="col" | NGI/RC <br />
! scope="col" | GOCDB Name <br />
! scope="col" | Contact <br />
! scope="col" | Service Type <br />
! scope="col" | Services and Resources <br />
! scope="col" | Preferred Payment Direct/Indirect <br />
! scope="col" | Price [1]<br />
|-<br />
! scope="row" | Albert Einstein Center, Univ. of Bern <br />
| UNIBE-LHEP <br />
| <br />
| Batch via CENTOS 5/6 ARC CE <br> <br />
| <br />
*500 cores (approx. 2 months in 2013) on highest priority; <br />
*Consumption preferably in one go, but distributed in time also possible <br />
*10-20 TB disk on local SE for consumption period via DPM SRM interface plus two weeks<br />
<br />
| Both <br />
| €0.03/hr; Storage free<br />
|-<br />
! scope="row" | NGI_BG<br />
|<br />
BG01-IPP<br />
BG08-MADARA<br />
| <br />
|<br />
Batch & Grid EMI-2/EMI-3 <br /><br />
Cloud - OpenStack - FedCloud<br />
|<br />
* 16 NVIDIA Tesla M2090 6 GB + 48 core Intel Xeon E5679 @ 2.53GHz with 4 GB of RAM per core<br />
* 576 cores Intel Xeon X5560 @ 2.8GHz with 1.5 GB of RAM per core<br />
* 800 cores Intel Xeon E5520 @ 2.26GHz with 1 GB of RAM per core<br />
* MPI jobs<br />
* Non-blocking InfiniBand interconnect<br />
* 240 TB of storage provisioned through EMI dCache, EMI DPM and Lustre FS /home and /scratch systems, OpenStack Swift and OpenStack Cinder<br />
| Both<br />
|<br />
€1.00/GPU/hour + 20% VAT (GPU-enabled servers) <br /><br />
€0.07/core/hour + 20% VAT (CPU only Intel Xeon X5560-based cluster) <br /><br />
€0.06/core/hour + 20% VAT (CPU only Intel Xeon E5520-based cluster)<br />
|-<br />
! scope="row" | CESGA-IBERGRID <br />
| CESGA <br />
| <br />
| Batch + Cloud <br />
| <br />
*720 cores (Batch) AMD Opteron(tm) Processor 6174 2.2Ghz <br />
*240 cores (Cloud) Intel(R) Xeon(R) CPU E5520 2.27GHz <br />
*MPI jobs <br />
*RAM from 4GB to 32GB per core<br />
<br />
| Direct <br />
| <br />
€0.02/core/HEPSPEC06 hour <br />
<br />
|-<br />
! scope="row" | IFCA-IBERGRID <br />
| TBD <br />
| <br />
| Batch (SLURM, SGE) or Cloud (FedCloud soon) <br />
| <br />
*Up to 2500 cores in servers with 4, 8, 16, 40 physical cores <br />
*Infiniband access to other nodes and storage possible for MPI jobs <br />
*RAM from 4GB to 256GB per core (max 1TB)<br />
<br />
| Direct <br />
| <br />
€0.05/core/hour for usual instances <br />
<br />
<span style="line-height: 1.5em;">€0.04/core/hour for multicore use (&gt;128 cores)</span> <br />
<br />
<span style="line-height: 1.5em;">Minimal storage free, requests up to 100 TB high performance storage possible</span> <br />
<br />
|-<br />
! scope="row" | Institute of Informatics of the Slovak Academy of Sciences <br />
| IISAS-FedCloud <br />
| <br />
| Cloud <br />
| <br />
*TBC<br />
<br />
| TBC <br />
| TBC<br />
|-<br />
! scope="row" | Latvian Grid <br />
| IMCSUL <br />
| <br />
| TBC <br />
| <br />
*TBC<br />
<br />
| TBC <br />
| TBC<br />
|-<br />
! scope="row" | Leibniz SC Centre <br />
| LRZ <br />
| <br />
| Globus <br />
| <br />
*Linux Cluster through Globus: GRAM5, GridFTP, Gsissh<br />
<br />
| Indirect <br />
| Price and capacity to be detailed later<br />
|-<br />
! scope="row" | MASTER-UP Srl <br />
| TBC <br />
| <br />
| EMI1/2, CEs, WNs, SE <br />
| <br />
*EMI1 middleware (phasing to EMI2) - CE's, WN's and SE. <br />
*10 nodes to start (intel 4core, ethernet connection) <br />
*Long experience in Molecular Science and Comp Chem, can provide support for such innovative applications and in the field of R&amp;D for combustion, energy and material science. <br />
*Managing a computer farm of 200 cores and 4TB of storage part of those resources are supporting EGI.<br />
<br />
| Direct <br />
| To be evaluated<br />
|-<br />
! scope="row" | NGI-DE <br />
| TBC <br />
| <br />
| UNICORE <br />
| <br />
*Cluster via UNICORE <br />
*Others can be offered (TBD)<br />
<br />
| Both <br />
| €0.06/hr + 19% VAT<br />
|-<br />
! scope="row" | NGI_GRNET <br />
| All HG-* <br />
| <br />
| Cloud and Grid <br />
| <br />
*TBC<br />
<br />
| TBC <br />
| €0.05/hr + 23% VAT<br />
|-<br />
! scope="row" | NGI-IT <br />
| <br />
INFN-BARI <br />
<br />
PRISMA-INFN-BARI (cloud) <br />
<br />
INFN-XX <br />
<br />
| <br />
| EMI, OpenStack <br />
| <br />
*gLite MW stack: Cream CE, LFC, Storm SE, WMS+LB, BDII. <br />
*Cloud: OpenSTACK<br />
<br />
| TBD <br />
| <br />
€0.07/hr + 22% VAT(grid) <br />
<br />
TDB (cloud) <br />
<br />
|-<br />
! scope="row" | PL-Grid <br />
| TBC <br />
| <br />
| Cloud + Batch via Cream / EMI <br />
| <br />
*Computing power, storage, and human support <br />
*Batch processing via Cream CE/EMI <br />
*Cloud access <br />
*Non-grid, local access to queue system<br />
<br />
| Both <br />
| Capacity and price are a subject to negotiation<br />
|-<br />
! scope="row" | TUBITAK ULAKBIM <br />
| TR-10-ULAKBIM <br />
| <br />
| Cloud + Batch <br />
| <br />
*Batch processing via Cream CE <br />
*VM Management using OpenNebula/OCCI <br />
*Infiniband connected Linux computing resources<br />
<br />
| Both <br />
| €0.15/hour 8 core 24 GB instance<br />
|-<br />
! scope="row" | UIIP NASB <br />
| BY-UIIP <br />
| <br />
| Cloud + Batch <br />
| <br />
*Batch processing via UNICORE or Cream CE <br />
*Intel Xeon E5472 (3,0 GHz, 4 Cores) 50 nodes / 100 physical CPUs <br />
*400 virtual CPUs <br />
*400 Gb memory <br />
*8TB Disk storage – 9-12TB Storage servers <br />
*Infiniband DDR 20Gbit/sec]<br />
<br />
| Indirect <br />
| $0.077/hr (TBD) 1 Core<br />
|}<br />
<br />
<br><br />
<br />
= Price in Accounting<br> =<br />
<br />
== Overview ==<br />
<br />
To test the mechanism for turning your accounting into a monetary charge charging rate needs to be defined. During the pilot this will be one number for your site in euros/HS06 hour. The number does not have to be real but obviously a ballpark figure would be good as it would give more realistic pilot reports. <br />
<br />
To define the charging rate the plan is to use a new feature of GOCDB which is known as ‘[https://wiki.egi.eu/wiki/GOCDB/Release4/Development/ExtensibilityMechanism Key-Value Pair Property Bag Extensiblity Mechanism]’. This allows the definition of arbitrary key-value pairs for a site or any service. If a community agrees a naming scheme for the keys and semantics for the values then they have a flexible and low overhead method of adding more information about anything defined in GOCDB. For anyone interested about more details this is documented [https://wiki.egi.eu/wiki/GOCDB/Release4/Development/ExtensibilityMechanism here]. <br />
<br />
== Instructions ==<br />
<br />
*If you do not have access rights to change features of your site please ask the site manager or a site admin to do it for you.<br />
<br />
#Go to the test GOCDB [https://gocdb-test.esc.rl.ac.uk/v5 https://gocdb-test.esc.rl.ac.uk/v5] which contains a copy of the production database. The same access rights apply but you will not change anything in the production database. <br />
#Navigate to your site. (My Sites) at the top of the left hand sidebar. <br />
#Just above the list of Services you should see a new box called 'Site Extension Properties' <br />
#At the bottom of this box there is a big green plus titled 'Add Properties' – click on it <br />
#This should show a window with two fields 'Property Name' and 'Property Value' <br />
#For 'Property Name' insert TEST_CHARGE <br />
#For 'Property Value' insert a real number which will be interpreted as euros per HEPSPEC06 Hour. <br />
#Click the button 'Add Site Property' <br />
#If you go back to the site view you will see TEST_CHARGE listed. You can edit or delete it.<br />
<br />
Once you have done this we will query GOCDB to pull the values for the pilot sites and the accounting portal will join them with usage data to produce a charging report.</div>Dmghttps://wiki.egi.eu/w/index.php?title=VT_EGI_Pay-for-Use_Technical_Details&diff=64526VT EGI Pay-for-Use Technical Details2014-02-13T10:36:03Z<p>Dmg: </p>
<hr />
<div>{{Template:EGI-PayForUse-PoC_menubar}} {{TOC right}} <br />
<br />
The following table provides an overview of the technical details of the resource and technology providers for the PoC. <br />
<br />
Further below are instructions for adding pricing into the accounting mechanisms. <br />
<br />
[1] All prices provided are interpreted as Euro per HEPSPEC06 hour. <br />
<br />
= Resource Providers, Services and Prices =<br />
<br />
{| class="wikitable"<br />
|-<br />
! scope="col" | NGI/RC <br />
! scope="col" | GOCDB Name <br />
! scope="col" | Contact <br />
! scope="col" | Service Type <br />
! scope="col" | Services and Resources <br />
! scope="col" | Preferred Payment Direct/Indirect <br />
! scope="col" | Price [1]<br />
|-<br />
! scope="row" | Albert Einstein Center, Univ. of Bern <br />
| UNIBE-LHEP <br />
| <br />
| Batch via CENTOS 5/6 ARC CE <br> <br />
| <br />
*500 cores (approx. 2 months in 2013) on highest priority; <br />
*Consumption preferably in one go, but distributed in time also possible <br />
*10-20 TB disk on local SE for consumption period via DPM SRM interface plus two weeks<br />
<br />
| Both <br />
| €0.03/hr; Storage free<br />
|-<br />
! scope="row" | NGI_BG<br />
| BG01-IPP<br />
| <br />
| TBC <br />
| <br />
*TBC<br />
<br />
| TBC <br />
| TBC<br />
|-<br />
! scope="row" | CESGA-IBERGRID <br />
| CESGA <br />
| <br />
| Batch + Cloud <br />
| <br />
*720 cores (Batch) AMD Opteron(tm) Processor 6174 2.2Ghz <br />
*240 cores (Cloud) Intel(R) Xeon(R) CPU E5520 2.27GHz <br />
*MPI jobs <br />
*RAM from 4GB to 32GB per core<br />
<br />
| Direct <br />
| <br />
€0.02/core/HEPSPEC06 hour <br />
<br />
|-<br />
! scope="row" | IFCA-IBERGRID <br />
| TBD <br />
| <br />
| Batch (SLURM, SGE) or Cloud (FedCloud soon) <br />
| <br />
*Up to 2500 cores in servers with 4, 8, 16, 40 physical cores <br />
*Infiniband access to other nodes and storage possible for MPI jobs <br />
*RAM from 4GB to 256GB per core (max 1TB)<br />
<br />
| Direct <br />
| <br />
€0.05/core/hour for usual instances <br />
<br />
<span style="line-height: 1.5em;">€0.04/core/hour for multicore use (&gt;128 cores)</span> <br />
<br />
<span style="line-height: 1.5em;">Minimal storage free, requests up to 100 TB high performance storage possible</span> <br />
<br />
|-<br />
! scope="row" | Institute of Informatics of the Slovak Academy of Sciences <br />
| IISAS-FedCloud <br />
| <br />
| Cloud <br />
| <br />
*TBC<br />
<br />
| TBC <br />
| TBC<br />
|-<br />
! scope="row" | Latvian Grid <br />
| IMCSUL <br />
| <br />
| TBC <br />
| <br />
*TBC<br />
<br />
| TBC <br />
| TBC<br />
|-<br />
! scope="row" | Leibniz SC Centre <br />
| LRZ <br />
| <br />
| Globus <br />
| <br />
*Linux Cluster through Globus: GRAM5, GridFTP, Gsissh<br />
<br />
| Indirect <br />
| Price and capacity to be detailed later<br />
|-<br />
! scope="row" | MASTER-UP Srl <br />
| TBC <br />
| <br />
| EMI1/2, CEs, WNs, SE <br />
| <br />
*EMI1 middleware (phasing to EMI2) - CE's, WN's and SE. <br />
*10 nodes to start (intel 4core, ethernet connection) <br />
*Long experience in Molecular Science and Comp Chem, can provide support for such innovative applications and in the field of R&amp;D for combustion, energy and material science. <br />
*Managing a computer farm of 200 cores and 4TB of storage part of those resources are supporting EGI.<br />
<br />
| Direct <br />
| To be evaluated<br />
|-<br />
! scope="row" | NGI-DE <br />
| TBC <br />
| <br />
| UNICORE <br />
| <br />
*Cluster via UNICORE <br />
*Others can be offered (TBD)<br />
<br />
| Both <br />
| €0.06/hr + 19% VAT<br />
|-<br />
! scope="row" | NGI_GRNET <br />
| All HG-* <br />
| <br />
| Cloud and Grid <br />
| <br />
*TBC<br />
<br />
| TBC <br />
| €0.05/hr + 23% VAT<br />
|-<br />
! scope="row" | NGI-IT <br />
| <br />
INFN-BARI <br />
<br />
PRISMA-INFN-BARI (cloud) <br />
<br />
INFN-XX <br />
<br />
| <br />
| EMI, OpenStack <br />
| <br />
*gLite MW stack: Cream CE, LFC, Storm SE, WMS+LB, BDII. <br />
*Cloud: OpenSTACK<br />
<br />
| TBD <br />
| <br />
€0.07/hr + 22% VAT(grid) <br />
<br />
TDB (cloud) <br />
<br />
|-<br />
! scope="row" | PL-Grid <br />
| TBC <br />
| <br />
| Cloud + Batch via Cream / EMI <br />
| <br />
*Computing power, storage, and human support <br />
*Batch processing via Cream CE/EMI <br />
*Cloud access <br />
*Non-grid, local access to queue system<br />
<br />
| Both <br />
| Capacity and price are a subject to negotiation<br />
|-<br />
! scope="row" | TUBITAK ULAKBIM <br />
| TR-10-ULAKBIM <br />
| <br />
| Cloud + Batch <br />
| <br />
*Batch processing via Cream CE <br />
*VM Management using OpenNebula/OCCI <br />
*Infiniband connected Linux computing resources<br />
<br />
| Both <br />
| €0.15/hour 8 core 24 GB instance<br />
|-<br />
! scope="row" | UIIP NASB <br />
| BY-UIIP <br />
| <br />
| Cloud + Batch <br />
| <br />
*Batch processing via UNICORE or Cream CE <br />
*Intel Xeon E5472 (3,0 GHz, 4 Cores) 50 nodes / 100 physical CPUs <br />
*400 virtual CPUs <br />
*400 Gb memory <br />
*8TB Disk storage – 9-12TB Storage servers <br />
*Infiniband DDR 20Gbit/sec]<br />
<br />
| Indirect <br />
| $0.077/hr (TBD) 1 Core<br />
|}<br />
<br />
<br><br />
<br />
= Price in Accounting<br> =<br />
<br />
== Overview ==<br />
<br />
To test the mechanism for turning your accounting into a monetary charge charging rate needs to be defined. During the pilot this will be one number for your site in euros/HS06 hour. The number does not have to be real but obviously a ballpark figure would be good as it would give more realistic pilot reports. <br />
<br />
To define the charging rate the plan is to use a new feature of GOCDB which is known as ‘[https://wiki.egi.eu/wiki/GOCDB/Release4/Development/ExtensibilityMechanism Key-Value Pair Property Bag Extensiblity Mechanism]’. This allows the definition of arbitrary key-value pairs for a site or any service. If a community agrees a naming scheme for the keys and semantics for the values then they have a flexible and low overhead method of adding more information about anything defined in GOCDB. For anyone interested about more details this is documented [https://wiki.egi.eu/wiki/GOCDB/Release4/Development/ExtensibilityMechanism here]. <br />
<br />
== Instructions ==<br />
<br />
*If you do not have access rights to change features of your site please ask the site manager or a site admin to do it for you.<br />
<br />
#Go to the test GOCDB [https://gocdb-test.esc.rl.ac.uk/v5 https://gocdb-test.esc.rl.ac.uk/v5] which contains a copy of the production database. The same access rights apply but you will not change anything in the production database. <br />
#Navigate to your site. (My Sites) at the top of the left hand sidebar. <br />
#Just above the list of Services you should see a new box called 'Site Extension Properties' <br />
#At the bottom of this box there is a big green plus titled 'Add Properties' – click on it <br />
#This should show a window with two fields 'Property Name' and 'Property Value' <br />
#For 'Property Name' insert TEST_CHARGE <br />
#For 'Property Value' insert a real number which will be interpreted as euros per HEPSPEC06 Hour. <br />
#Click the button 'Add Site Property' <br />
#If you go back to the site view you will see TEST_CHARGE listed. You can edit or delete it.<br />
<br />
Once you have done this we will query GOCDB to pull the values for the pilot sites and the accounting portal will join them with usage data to produce a charging report.</div>Dmg