RE: Backup and Compare Performance ... lots of small files
Title: RE: Backup and Compare Performance ... lots of small files I'm not familiar with this particular backup device but it sounds like the read head is out of alignment with the write head. To eliminate other causes we would 1. clean the heads 2. run a short test backup/ compare/ restore on a fresh tape. If the problem remains, get the heads checked. Rob Davies TSG-NDSA (Technical Services Group - Non Datacentre Server Admin) CWO -Original Message- From: Church Initiative WebMaster [mailto:[EMAIL PROTECTED]] Sent: Friday, 19 May 2000 5:19 To: retro-talk Subject: Re: Backup and Compare Performance ... lots of small files I am having a related problem. Up until two days ago everything was working fine, but since then a backup has yet to complete. Last night, the backup computer did not even finish backing up itself. Our back up server is a G3 400 Mhz with 192 MB of RAM with 23 GB of Hard drive space and our backup device is a VXA 66 GB tape drive. After running for 13.5 hours, the backup had managed to backup 12.4 GB but only compare 10.5 GB. The MB/min for the copy was 130.4 and for some unknown reason plummeted to only 15 MB/min for the compare! Remember this is NOT over a network, but on the local machine. There are only 153 files that make up the 12.4 GB so why the slowdown Stiles Watson Director of Information Services The Church Initiative, Inc. www.divorcecare.org www.griefshare.org - Original Message - From: John F. Lambert [EMAIL PROTECTED] To: retro-talk [EMAIL PROTECTED] Sent: Wednesday, May 17, 2000 7:45 PM Subject: Re: Backup and Compare Performance ... lots of small files Anyone had similar experiences where the hierarchy of the data has a dramatic effect on performance? I find that Retrospect on a 10base T network, Backup Server on a Pentium II 233, Client on a PowerBook G3 266, (or even better, on Virtual PC on a PB G3 266) can achieve 60MB/min transfer rates (= 1MB/sec = pretty much the maximum transfer rate on 10BaseT) when transferring large files, but performance drops away to around 12MB/min when reading lots of small files - I am surprised by this large drop in performance too! John -- -- To subscribe: [EMAIL PROTECTED] To unsubscribe: [EMAIL PROTECTED] Archives: http://list.working-dogs.com/lists/retro-talk/ Problems?: [EMAIL PROTECTED] _ NetZero - Defenders of the Free World Click here for FREE Internet Access and Email http://www.netzero.net/download/index.html -- -- To subscribe: [EMAIL PROTECTED] To unsubscribe: [EMAIL PROTECTED] Archives: http://list.working-dogs.com/lists/retro-talk/ Problems?: [EMAIL PROTECTED]
Re: Backup and Compare Performance ... lots of small files
Anyone had similar experiences where the hierarchy of the data has a dramatic effect on performance? I find that Retrospect on a 10base T network, Backup Server on a Pentium II 233, Client on a PowerBook G3 266, (or even better, on Virtual PC on a PB G3 266) can achieve 60MB/min transfer rates (= 1MB/sec = pretty much the maximum transfer rate on 10BaseT) when transferring large files, but performance drops away to around 12MB/min when reading lots of small files - I am surprised by this large drop in performance too! John -- -- To subscribe:[EMAIL PROTECTED] To unsubscribe: [EMAIL PROTECTED] Archives:http://list.working-dogs.com/lists/retro-talk/ Problems?: [EMAIL PROTECTED]
Backup and Compare Performance ... lots of small files
Guys, I've just installed a new server for a client and all was going well until the client stated using his existing data which we had copied to a group share ... its a slug. This is particularly noticable when doing a backup, copy is bad but it is even worse during the compare phase. Originally I thought there was a network or hardware problem, but through a process of elimination (inc swapping the server) I conclude its SAMBA or possibly the way the user has structured his data. Having a closer look at the data, it would appear that the user has some directories which contain 1300 small files. Results from this afternoon were 24.2 Mb/minute copy, 5.6 Mb/minute compare to a file based storage set. The test configuration consisted of the server (Qube2) and the backup workstation (Pentium III 550) connected to a 100baseT switch ... both machines were 100Mbps Full Duplex. I don't consider these figures to be anything to boast about. Anyone had similar experiences where the hierarchy of the data has a dramatic effect on performance? I have used this configuration on other sites without such problems ... the only significant difference is the clients data. I guess the best solution is to suggest to the client that he structure his data a little better than stuff everything in a single folder. I would be interested to know why the performance of the compare phase is so bad. Restore is equally as bad, but I guess that is to be expected as I gather compare is a restore without the actual "write to target". Cheers, Malcolm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Information Alchemy Pty Ltd ACN 089 239 305 Canberra, Australia Malcolm McLeary Mobile: 0412 636 086 Managing DirectorEmail: [EMAIL PROTECTED] This message was sent using Claris Emailer 2.0v3 for Macintosh. -- -- To subscribe:[EMAIL PROTECTED] To unsubscribe: [EMAIL PROTECTED] Archives:http://list.working-dogs.com/lists/retro-talk/ Problems?: [EMAIL PROTECTED]