Please read these instructions before posting any event on Fermilab Indico

Indico will be unavailable on Monday Feb 17th from 4:30-5PM CST due to server maintenance.

Storage Access Architecture

US/Central
Description
Storage Access Architecture (20 Oct 2015)

   * attendance: Burt, Gene, Andrew, Stu, Margaret, OLI
 
   * went through slides attached to indico agenda
   * in short: store configuration files as blobs in either Postgres or Couchbase and serve it out through engineX http caches like conditions
   * number of files:
      * mu2e: ~4-5E6
      * NOvA: ~1E3
   * experiments are using SAM projects for accounting of their workflows
      * interaction rates needed: 10k-100k concurrent accesses
   * development effort
      * reuses a lot of components
      * no modifications to SAM
      * minor modifications to ifdh -> Marc estimates one afternoon development and testing
      * slight modifications to webserver frontend to access new backend
      * new backend
         * Postgres is straight forward -> Igor estimates 3 days for development and testing
         * Couchbase takes longer -> no estimate
            * talked to Mitch about Couchbase, they are positive, they envision that they have to do this (or similar nosql db technology?) anyway
            * Couchbase was already used for Big Data studies, there is already experience
   * what about data safety
      * experiments stored their files on tape for long term storage
      * database under backup would be equally acceptable, like for conditions
   * there are examples that this works, this could be a generic solution for any kind of blob access from jobs
   * tests
      * Couchbase achieved 10E6 transactions per second
      * Postgres: unknown the blob transaction rate
   * user interface plan:
      * users get ifdh mode to post configurations/blobs to DB
   * migration
      * straight forward using tools to ingest data to DB and SAM replication, could be lengthy for mu2e with many million files
      
      
   * Architecture
      * add new storage location: blob DB to read support files
      * support files will be enabled for this storage location
   * need decision whether to remove support files from non-tape-backed storage
      * depends on how big effort is to create new DB for blobs, if small, remove
 
   * action items
      * Andrew: answer questions:
         * Postgres blob size limitation and scaling estimates
         * Overhead to create new blob DB
      * Andrew and Margaret
         * estimate effort to run this new service in production
         * schedule SPPM outside the usual schedule this or next week to get sign off from Panagiotis
      * OLI
         * modify storage access architecture with new input
 
There are minutes attached to this event. Show them.
    • 09:00 09:30
      Configuration file handling 30m
      Speaker: Dr Andrew Norman (Fermilab)
      Slides
    • 09:30 10:00
      Discussion 30m