Steve,
It is going directly to tape. Our backups and backups of our redo log
files go straight to tape.
Raminder Braich
> -----Original Message-----
> From: Steve Schaub [SMTP:Steve.Schaub AT HAWORTH DOT COM]
> Sent: Friday, January 26, 2001 3:27 PM
> To: ADSM-L AT VM.MARIST DOT EDU
> Subject: Re: Achievable throughput using Gigabit connections?
>
> Raminder,
> If I read this right you are seeing 17mb/sec sustained in a single backup
> session - correct? Is this going direct to tape, or are you using a disk
> pool?
>
> Steve Schaub
> Haworth, Inc
> email: Steve.Schaub AT Haworthinc DOT com
>
> >>> raminder.braich AT DAVEY DOT COM 01/26 3:03 PM >>>
> Steve
> Our Environment
> TSM Server: Netfinity 5500R 2 processors, 1 GB RAM, Giga bit,
> WIN200, TSM4.1
> CLient: NEtfinity 7000 4 processors, 1GB RAM, WIN NT, ORACLE7.3.4 ~
> 360GB TSM4.1
> Cisco network switch.
> Our backup time is a litlle over 6 hours approx 17 MB/sec. Previously on
> TSM 3.7 we used to see max 11MB/sec but with TSM 4.1 and WIN2000 it is 17M
> /sec.
>
> Raminder
>
> > -----Original Message-----
> > From: Mills, John [SMTP:John_Mills AT PURINAMILLS DOT COM]
> > Sent: Friday, January 26, 2001 2:31 PM
> > To: ADSM-L AT VM.MARIST DOT EDU
> > Subject: Re: Achievable throughput using Gigabit connections?
> >
> > Steve,
> >
> > We have seen our database restore reach speeds of about 25 MB/s
> > across a gigabit connection, but a more realistic average was 12-15MB/s.
> > 80 GB/hr is about 23 MB/s, and we could see that if our disk were
> striped
> > in an ideal situation.
> >
> > John
> >
> > > -----Original Message-----
> > > From: Steve Schaub [SMTP:Steve.Schaub AT HAWORTH DOT COM]
> > > Sent: Friday, January 26, 2001 8:11 AM
> > > To: ADSM-L AT VM.MARIST DOT EDU
> > > Subject: Achievable throughput using Gigabit connections?
> > >
> > > I am looking to find out what kind of mb/sec other shops are getting.
> > Our
> > > environment is:
> > > AIX 4.3.3 Server (3466-C00 = H50 2cpu, 1gb mem, 2-gbit ethernet,
> 1-100mb
> > > ethernet) running TSM 3.7.3.8
> > > 3494 library using 3590E drives with 3590K tapes
> > > Switch is Cisco 6509 on a dedicated vlan segment
> > > Client is HP V2200 (16cpu, 12gb mem, 1gbit ethernet dedicated to
> backup
> > > vlan) running TSM 3.7.2 under HP-UX 11.0
> > >
> > > Specifically, when we try to archive a 100gb Oracle database on the HP
> > (as
> > > a file system with approx 450 files, not using EBU/RMAN, residing on
> an
> > > EMC 3700 Symmetrix), we are seeing aggregate throughput in the
> 6-7mb/sec
> > > range, which translates into approx 4.5hrs. When we watch the Aix
> > > monitor, the gbit interface varies from 15 down to 1.5mb/sec. The cpu
> > > never goes over 20%. The consultant who helped us get this installed
> &
> > > setup claims that a business down the street with almost identical
> > > hardware/software is able to get a single backup stream to do 80gb/hr.
> > So
> > > far we have involved HP, Tivoli, Aix, Cisco, EMC and almost every
> techie
> > > in our shop without being able to find a conclusive answer as to where
> > our
> > > bottleneck is.
> > >
> > > Are other shops seeing anything close to 80gb/hr in a single stream
> > using
> > > gb to gb ethernet connections? Any pointers as to where to look, or
> > > recommended tools for at least diagnosing the problem? Thanks in
> > advance.
> > >
> > > Steve Schaub
> > > Haworth, Inc
> > > email: Steve.Schaub AT Haworthinc DOT com
|