Difference between revisions of "Top-BDII list for NGI"
(→Top-BDIIs operated by the NGIs)
(→Top-BDIIs operated by the NGIs)
|Line 94:||Line 94:|
Revision as of 16:01, 1 June 2012
|Main||EGI.eu operations services||Support||Documentation||Tools||Activities||Performance||Technology||Catch-all Services||Resource Allocation||Security|
top-BDII Availability and Reliability
This page contains the list of the Top-BDII instances that are operated by the NGIs. Starting from October 2011, these instances will be considered for the generation of the monthly NGI Availability and Reliability report.
Availability and Reliability figures are extracted from MyEGI (ROC profile, OPS VO).
- if you think that a wrong top-BDII entry is mentioned in the table below, please send a GGUS ticket to the Operations support unit indicating the entry to replace and the new service end-point.
- the same top-BDII instance can be reported for more than one NGI. This is the case if two ore more NGIs are sharing this service, this is typically the case when they are operated by the same Operations Centre.
- several top-BDII instances can be reported for one NGI. This is the case one a NGI deploys failover at client side using multiple top-BDII instances as alternative contact points (MAN05). In this case the NGI overall top-BDII hourly availability is calculated by OR-ing the hourly availability of the individual top-BDIIs.
Top-BDIIs operated by the NGIs
|Operations Centre||Top-BDII host(s)||Notes|
|Austria||egee-bdii.cnaf.infn.it||Austrian sites are collaborating with the NGI_IT Operations Center|
|Czech Republic||bdii1rr.farm.particle.cz||DNS alias for two instances running at different sites.|
|Spain||topbdii.egi.cesga.es, gridii01.ifca.es, bdii.pic.es, topbdii01.ncg.ingrid.pt||Spanish sites also point to the Portuguese TopBDII|
|Finland||No Top-BDII currently in use|
|Ireland||All sites use the CERN top-bdii|
|Italy||egee-bdii.cnaf.infn.it||five instances under dns round robin|
|Norway||No Top-BDII currently in use|
|Poland||bdii-top.reef.man.poznan.pl, zeus60.cyf-kr.edu.pl, topbdii.polgrid.pl||All 3 working under DNS pool bdii.cyf-kr.edu.pl|
|Portugal||topbdii01.ncg.ingrid.pt, topbdii.egi.cesga.es, gridii01.ifca.es, bdii.pic.es||Portuguese sites also point to the Spanish TopBDIIs|
top-BDII availability/reliability statistics are based on availability/reliability data that is published by MyEGI through the SAM Programmatic Interface (SAM PI). MyEGI is the authoritative source of availability/reliability data (GridView is now decommissioned).
A brief description of the algorithm adopted to compute top-BDII availbility and reliability statistics follows.
Different schenarios are possible.
Scenario 1. Single top-BDII
The NGI reports only one Top-BDII. It could be a single top-bdii instance, or alternatively a DNS alias for a pool of top-BDII instances (the DNS alias MUST BE monitored by the NGI Nagios). In this case Availability and Reliability are the monthly Availability and Reliability reported directly by the MyEGI Programmatic Interface.
Note: If the alias is used by the sites but only the single instances in the pool are monitored by Nagios, then in the table above the list of instances per NGI has to be reported. In this case Availability and Reliability are computed according to Scenario 2.
Scenario 2. List of top-BDIIs
The NGI reports a list of top-BDIIs. These are all used by the sites to configure failover at client side. In this case the algorithm steps are the following:
- Query MyEGI to get the hourly A/R values for every top-BDII instance in the list. (Example of query)
- Hourly Availability and Reliability: given the list of Availability/Reliability figures for the top-BDII in the list, the maximum Availability/Reliability figures are computed. These are selected to be the hourly Availability/Reliability. (Example of query)
- Monthly Availability and Reliability: the monthly Availability/Reliability is the arithmetic mean of the hourly Availability/Reliability figures in the reference month:
- Monthly A/R: (sum of the Hourly A/R figures)/(number of hours when the status was known).