tpmsph@ecsvax.UUCP (Thomas P. Morris) (04/26/88)
The University of North Carolina at Chapel Hill, School of Public Health, Division of Computing and Information Services, runs a 3-node VAXcluster, attached to which we have 4 RA81's. One is a system disk, two are nearly full of user directories and data, and one has user utilities, alternate page and swap files, and some user directories and data. Our primary system usage is for support of office automation and word processing usage for eleven departments or divisions of the School. To provide user file safety, we perform: - daily incremental backups - weekly incremental backups - weekly HSC disk image backups - monthly physical disk image (standalone) backups The daily and weekly incrementals are done on only 3 of the disks. The HSCs are done for all disks, as are the monthly standalones. Users are allowed to remain logged on for the daily backups, but the system is essentially shut down for the weekly and monthly backups, for ~4-5 hours for the weekly, and 8-10 hours for the monthly. Yes, it seems to be overkill to us, too. There has to be an easier way! How do other sites handle this? We'd like to rotate the "monthly" backups to one disk / week. We've only got 1 TA78 tape drive: would a second tape drive help? What about reserving a new (additional) RA81 to be used as follows: mount it as a shadow volume, tahen take the shadow offline, and perform an HSC backup on that? That could be done 1 disk at a time, one week at a time, with users "locked out" only for the duration of having the shadow volume built. Comments? Suggestions? Thanx. ----------------------------------------------------------------------------- Tom Morris BITNET: TOM@UNCSPHVX UNC School of Public Health UUCP : ...!mcnc!ecsvax!tpmsph ----------------------------------------------------------------------------- -- ----------------------------------------------------------------------------- Tom Morris BITNET: TOM@UNCSPHVX UNC School of Public Health UUCP : ...!mcnc!ecsvax!tpmsph -----------------------------------------------------------------------------
nagy%warner.hepnet@LBL.GOV (Frank J. Nagy, VAX Wizard & Guru) (06/29/88)
The Fermilab Research Division Electrical/Electronics Department Controls Group operates a Local Area VAXCluster consisting of two MicroVAXes and (currently) 6 VAXStations. There are currently 4 RA81s consisting of a system disk and two disks for user files and data (we are a software and hardware development environment). The 4th RA81's file have been moved or unloaded in order to turn it into a second system disk since we are converting our LAVC to dual boot nodes in order to support growth from 8 nodes to 23-24 nodes in the next month. We currently have about 1 GByte of active files on these 4 RA81s. We plan to expand the file system this year by at least 50%. Backup is done with a pair of DigiData Gigastores connected to the two MicroVAXes. These are helical scan tape systems using VHS tape cartridges to store up to 2.5 GBytes each. All backups are done online with the systems live by batch jobs. The current schedules are: - Every night do a /IMAGE backup of all disks. Currently this is done with a pair of RA81s to each Gigastore. The system manager comes in on weekday mornings and changes out the cartridges. The backup jobs run at 3 AM and take about 1.5 hours. - At 12:00, an incremental backup is done of the entire file system onto a TK50 on the boot node. - At 18:00, an incremental backup is done of the entire file system onto a TK50 on the other MicroVAX node. The incremental backups are just fluff (to some extent), but have helped when someone has deleted a file in the past. Our basic backup philosophy is "hands off"; we have no operators to mount tapes, so we have acquired equipment so we don't need same. Doing so allowed us to change our backup schedules. ORIGINALLY, we did backups to TK50s. A weekly /IMAGE backup was done (again online with users on the system) Monday morning and afternoon. Incrementals were done nightly by a 3 AM batch job and the TK50s changed on weekday mornings. Things are soo much easier now with the Gigastores. = Frank J. Nagy "VAX Guru & Wizard" = Fermilab Research Division EED/Controls = HEPNET: WARNER::NAGY (43198::NAGY) or FNAL::NAGY (43009::NAGY) = BitNet: NAGY@FNAL = USnail: Fermilab POB 500 MS/220 Batavia, IL 60510