RAL Tier1 Network Infrastructure

From GridPP Wiki
Revision as of 11:59, 28 November 2007 by Andrew sansum (Talk | contribs)

(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to: navigation, search

This is a Wiki area to Tier-1 Network Meeting Actions

Action ID prefix Status
N = From Tier-1 Network Meeting Open = Action has been created
T = Added by Team members Progress = Action is being worked on
P = Created by other project members Closed = Action is complete
Rejected = Action is rejected


Action ID Priority Owner Action Title Target date Status Date closed Notes


N-05 Medium Seelig Permanent Installation Subnet 21/12/07 Open Subnet for installation of clusters - external to RAL RAS will check with E-Science for requirement. [21/03/07]Further action not required until next procurement delivery is pending.[16/5/07]Installation subnet will be needed by end of September for October delivery.[19/09/07]Now needed by 1st December[28/11/07]Delivery is scheduled for W/B 7th January. Installation network must be

in place before Christmas - upgraded to Medium


N-07 High Seelig Automated/Delegated DNS changes 31/03/07 New We need a way of automatically updating DNS records. Has become more urgent. We will talk to Chris offline. [21/03/07]Need to arrange meeting with RAS, CS, MJB.[16./5/07]Meeting happened but ran out of time and did not discuss. new meeting will be scheduled by RAS.[11/07/07]A meeting is arranged late in the month[25/7/07]Meeting held - plan to try to use network access to an existing mysql database. RAS will find a Tier-1 team member to work on this.[19/09/07]Matt and Derek will work on this.


N-10 Low Bly May have a requirement for jumbo frames New Martin will mail Robin with our requirement. [11/07/07]Martin has not yet arranged a meeting.[03/10/07]Discussion with CS indicates that all RAL equipment is capable of handing jumbo frames.
N-11 High Bly Tier-1 requires more substantial network switch 1/3/07 New [31/10/07] Paper on T1 needs written, RAS has seen. MJB will send to Robin, Chris and Dave J. [28/11/07]A document was circulated and RAS agreed to schedule a meeting to discuss. Need this switch paid for before the end of the FY. Preferably needs to be operational

by early March.


Closed Actions

Action ID Priority Owner Action Title Target date Status Date closed Notes
N-01 High Tasker Tier-1 Connected to SJ5 at 10Gb 30/04/07 Closed 31/10/07 Scheduled date is now mid 1Q07. This action is a reminder to continue monitoring the status. Procurement is going ahead and is on track for new planned date[7/02/07]UKERNA report that they expect to provide this at the end of March. Chris expect 1 month to provide the full bandwidth to the Tier-1, but initial connection will be within a week of UKERNA providing the link.[7/3/07]UKERNA may not be able to provide both links but hopefully we should have one link looped back by 23rd March. Robin has also asked for a status update but has not yet received a reply.[21/03/07] 10Gb to London ready, 10G to reading imminent, routing change planned for 27th if links are ready andf Janet notice is OK.[18/4/07]On 10th April 2*10Gb links brought online (connection/backup). On 17th April site changes were made to connect Router A 2Gb to Firewall, but problems delayed 2*1Gb from router B.Firewall to SAR is now 2*1Gb, plans to move to 4*1Gb on 24th April. Chris is concerned as to if everything is working OK. We looked at traffic rates but concluded that we needed to wait 2 weeks and review statistics at the next meeting.[2/5/07]gridmon rates were good (>200Mb/s tcp single stream) until Tier-1 traffic turned on after which rates dropped to 40-50Mb/s.[16/5/07]Link from SAR to UK will go in place this week, then filters will be required testing is required. Chris will circulate a planned schedule later today.[13/06/07]Tier-1 has provided address details to Chris for routing purposes. A router card has been installed but other work has been delayed because of technical problems with site routing.The 10Gb bypass was installed but had to be removed after problems occurred (change backed out). It will be re-inserted next Tuesday. We are at least 3 weeks away from having this in production.[11/07/07]Chris expects to have the firewall ruleset implemented by the end of the week and Tier-1 can start testing next week.[25/07/07] Has not been able to implement our ruleset. Tier-1 expects to hit main site network any day now.[22/8/07]No progress [19/09/07]Chris has a test network installed and now has to replicate and test the filters. Expects to be able to implement this next week in coordination with the Tier-1. Phase I - Implement physical bypass and firewall rule set. Phase 2 - designate a test host on Tier-1 and test routing. Phase III identify a CASTOR disk pool of about 6 servers and route via bypass. Phase IV Route 179/24 via bypass. Expect phase I and II next week. Phase III/IV ASAP. [03/10/07]Start with a few hosts as a test. [31/10/07] Bypass now running for selected disk pools, the rest to be migrated over time. Closed.
N-02 High Seelig Second 1Gb link available for Tier-1 20 February 2007 Closed 17/4/07 We require a date for availability of second 1Gb link. Will destination-route. Will need subnets for Tier-2 SRMS. Bundles at firewall/routers set up but not active. Will become so on Tuesday. Second 1Gb connection has IP addresses and is active (to test router). Will schedule for 28th November.This still requires BGP peering to be arranged arranged with UKERNA. We agreed that the latest this could be deployed before xmass was Tuesday 19th December.[7/02/07]Agreed that this was still required ASAP. RAS will provide a list of high bandwidth sites.[7/3/07]We have outbound traffic enabled on the second link but Chris does noy yet understand the load balancing. [21/03/07] Link now moot. Load balancing traffic in place on 2nd link from BAR outward. Propose action closed.Full 2Gb path all the way to SJ5 - action closed


N-03 Medium Seelig Site network throughput monitoring Closed We need throughput statistics for site link to SJ4. Chris is investigating. Robin will mail a URL to the JANET netsight. [13/12/06] Robin distributed a link to the UKERNA monitoring.[7/2/07]Testing a number of flow analysis products.[7/3/07]Martin now has Cacti access to router A. Networking has purchased a product which will be deployeyed once the new 10Gb in production.


N-04 Critical Seelig Permanent IP address in visitors network 15 November 2006 Closed 16 Nov 2006 Required to allow remote access to Streamline installation. Provided 16/11.
N-06 Medium Bly Requirement for configuration of network 184 Closed 17 Nov 2006 With advice from Nick Moore, I have reconfigured the 184 network connections in T1-land to provide access for the PPS systems.
N-08 High Seelig Arrange OPN peering for CMS 31/3/07 Closed 31/10/07 We need to be able to route CMS Tier-1 sites over the OPN [21/03/07] Waiting for an application for a public AS number to be approved.[18/4/07] Waiting for AS number from RIPE - expect it in 1 weeks time. If a problem we will route statically.[16/5/07]Seems to be a problem contacting UKERNA to pass on required information. Robin will inquire about how T1s are using the OPN. This is becoming very urgent as it will need to be in place before CMS start CSA07 in June.[13/06/07]AS number is approved in principle but we have to sort out existing registration paperwork. Plan is to have the registrations up to date this week. Initially static routing will be put in place to CNAF until AS number is set up.[11/07/07]We have still not resolved obtaining the AS number from ripe. RAS reported that this is now becoming very urgent as high rates are expected shortly.[25/7/07]We have had to obtain signoff at CEO level to update our contact details with RIPE. Fax expected to be sent today. Hope to get the AS number within next 4 days.[22/8/07]We have the AS number and Networking have set up the RAL end of the BGP peering. We have still to contact CERN to arrange the peering which is expected to be done in the next day or so. After this it will be up to the Tier-1 to bilaterally arange routing changes with its peers.[19/9/07]BGP advertising will commence at 11:00. CERN will not advertise these routes initially. [3/10/07] RAS to email T1s to obtain information about T1 what happens when BGP routing starts so that we do route to sites properly. CMS will stop T1-T1 transfers and recommission from scratch over OPN, to start with FermiLab. Opening position at RAL: not to route by default. See if it is possible to do it Thursday afternoon, otherwise Monday. CS to initiate discussion on OPN routing list concerning routing policies. c.f. N-12. [31/10/07] OPN peering now in place for 6 T1s and CERN. The others to be enabled as required. Closed.
N-09 High Seelig Plan and implement changes to internal pathing for T1 and site traffic Closed CS/MJB to look at plan for internal site pathing to allow T1 data traffic passed firewall at 10Gb/s.[18/04/07]We will meet and discuss schedule for bypass.[16/5/07]Superseeded by N-01
N-12 High Seelig OPN routing policy. 03/10/07 Closed 31/10/07 CS to raise issue of routing policy with OPN routing group. [31/10/07] Issue resolved without need to contact OPN group. Closed.