Download presentation
Presentation is loading. Please wait.
Published byDarcy Webb Modified over 9 years ago
1
LHCb input to DM and SM TEGs
2
Remarks to DM and SM TEGS Introduction m We have already provided some input during our dedicated session of the TEG m Here are a list of questions we want to address for each session o Not exhaustive, limited to one slide per session m It would have been useful to hear what are the proposals for evolution of SM and DM as well o These should be proposals, not diktat o Extensive discussions are needed with users and sites before embarking on full scale “prototypes” that no longer are prototypes m Whatever the future is, WLCG must ensure the long term support (no external “firm”) o Do not underestimate the amount of work needed for users to adapt o Therefore plan well ahead, gather functionality requirements… F2F Data and Storage Management TEG, Amsterdam2
3
Remarks to DM and SM TEGS F2F Data and Storage Management TEG, Amsterdam3 Data Placement and Federation m Does it imply replica catalogs are no longer needed? o How is brokering of jobs done? o Random or “where most files are” m When is data transfer performed? o By the WMS: implies a priori brokering, incompatible with pilot jobs o By the job: inefficiency of slots o Is it a cache (life time?) or just a download facility? m What is the advantage compared to placement using popularity? o Limited number of “master replicas” (e.g. 2) o Add replicas when popularity increases o Remove replicas when popularity decreases o … but still with a catalog and job brokering m What is the minimal number of sites for which it becomes worth it?
4
Remarks to DM and SM TEGS WAN protocols and FTS m We need third party transfer! o http 3 rd party transfer? OK if commercial, why support it ourselves? m We need a transfer “batch” system! o Asynchronous bulk files transfer m Whatever reliable and efficient underlying protocol is used is just fine… m There is a need to define the service class where the file has to be put (or one service class per SE) m What about the dedicated network (OPN)? o Requires a service for using it? o Not all bells and whistles may be necessary m The real point is for a user (experiment): o Transfer this list of LFNs to this SE (SE = storage class at site) P The actual physical source is irrelevant P The TS should discover whether there is an online replica, if not it should bring it online before making the transfer P Ideally it (or the SE) should register the new replica (keep consistency) m All this was already said in… Mumbai (February 2006)! m FTS 3 was looking promising… why is it dead? F2F Data and Storage Management TEG, Amsterdam4
5
Remarks to DM and SM TEGS Management of Catalogues and Namespaces m See Data placement… m Do we need a replica catalog? o LHCb answer is YES: we want to be able to do brokering of jobs o May only contain information on the SEs (+ file metadata + usability flags) m Do we need a catalog with URLs? o Not necessarily: the URL can be formed from the SE information and the LFN (trivial catalog), as SE information is quite static. m Do we need a single URL (used for transfers and for protocol access)? o No problem as long as the access is transparent and fast m See SRM slide for more comments… o Namespace vs storage class? F2F Data and Storage Management TEG, Amsterdam5
6
Remarks to DM and SM TEGS Security and Access Control m We MUST protect our data from deletion o LHCb doesn’t care about protecting from access so much m The current situation is INACCEPTABLE o Anyone with little knowledge (available on the web) can delete all files in Castor! o VOMS (or equivalent) identification and authorisation is a MUST! What about ARGUS? P Identity and role P Currently in Castor we have only 2 (uid, gid)! P Protection done by the LFC, but all backdoors are open o Backdoors should be closed (nsrm, stager_xxx active commands…) o Explicit “delete” permission would be desirable m Change of DN should be straightforward (not trivial, but OK in LFC, DPM) o Action from VO administrator F2F Data and Storage Management TEG, Amsterdam6
7
Remarks to DM and SM TEGS Separation of disk and tape m We need two (and only two) storage classes: o T1D0 and T0D1 o This is because no space (storage class) change is possible in some implementations of SRM m T1D0 has two functions: o Archive of T0D1 o Permanent storage of read-few data (RAW, RECO) P For this the BringOnline functionality is mandatory P We need to access the data directly from the MSS without a need for replication onto another storage P Pinning is also a must (suboptimal usage of tape drives without it) d Help to the garbage collector F2F Data and Storage Management TEG, Amsterdam7
8
Remarks to DM and SM TEGS Storage Interfaces: SRM and Clouds m Clouds??? m We need an interface for: o BringOnline o Pinning o Defining the storage class (unless different endpoints are used, i.e. different SEs) o Currently (Mumbai) this is done by gfal (what is its future?) m SRM is far from perfect but… o It provides the above o All efforts put into defining a standard were a miserable failure… don’t expect any other interface will be any better m … but… o We could probably wrap the minimal functionality on top of the SE native interface, if available P BringOnline and pinning not available for dCache except in SRM P Can xroot provide this functionality? P Isn’t there the danger it becomes as clumsy as SRM depending on the implementation? F2F Data and Storage Management TEG, Amsterdam8
9
Remarks to DM and SM TEGS Storage IO, LAN Protocols m What is wrong with POSIX file: protocol? o Very efficient in the StoRM-GPFS implementation used at CNAF o Of course abuse is could be a danger (recursive ls) but this could be taken into account in the implementation (throttling) o Almost anybody can now write a fuse plugin to make it happen, so why not use a powerful commercial protocol? m Should access protocols be more than protocols? o i.e. interact behind the scene with the MSS, discover the file location etc… o Can a tURL be just an access point? P ://diskserver.cern.ch// P … or better file: P Avoid accepting URLs like “/castor/cern.ch/…” d Needs to be fixed in application layer? No “guesses”? o Do we need different URLs for different operations? P Transfer and posix-like access F2F Data and Storage Management TEG, Amsterdam9
10
Remarks to DM and SM TEGS Conclusion m Whatever the future is: o Consider we need a permanently running system o No disruption of service of more than 24 hours for migration o Millions of replicas are in the current system and… it works (even when not optimal) o Any drastic change requires a lot of work on the user side (experiments framework) o Old and new systems must exist in parallel m Requirements may be different depending on the Computing Model of experiments o 7 to 12 analysis centers (LHCb) is different from 50 to 70 centers o Solutions may not be universal and complication may not be required F2F Data and Storage Management TEG, Amsterdam10
Similar presentations
© 2024 SlidePlayer.com. Inc.
All rights reserved.