Archives of the TeradataForum

Message Posted: Wed, 12 Apr 2006 @ 16:05:01 GMT

	<Prev	Next>		<<First	<Prev	Next>	Last>>

I have included the original notes from Andy and Hock below.

Our configuration (3-1-2) includes 12 nodes, 4 BAR servers, 8 tape drives (2 tapes per server). (NOTE: This may not be an ideal configuration; in prior levels of Teradata hardware/software, 2-1-2 was considered ideal.)

When all 8 tape drives are simultaneously spinning, throughput on 7 of the tapes is around 110 gb per hour. The slow drive is around 90 gb / hour. (Why????)

An interesting finding, one our jobs archived a 100 gb table (including index space) in 30 minutes. This was the only job running. Backing off the index space (estimated at 25 gb) translates into a tape that is spinning at a rate 150-160 gb / hour.

The system performs as advertised (against some conservative projections), but some tuning may be able to improve our throughput.

Rgrds,

Bill Gregg

-----Original Message-----
From: hock lim
Sent: Friday, March 17, 2006 14:00
Subject: Re: [TDATA-L] Experiences with TSM and Teradata

Andy Crockett wrote (7 Feb 2006):

Andy,

Teradata has a documentation detailing the use of Tivoli Access Module for TSM backup.

Basically the backup is still performed using the well-known arcmain utility but the data is sent via Tivoli Access Module then to the Storage Agent; the tapes are still managed by TSM Server. No magic here.

It is truly an open backup solution where almost anything goes.

However, something to look out for:

- Gigabit NIC tunables

- Need a private LAN

- No sharing of ASIC in switches

- Careful what device drivers to install for SAN

- Point to only one node per bar server to improve throughput

- How to schedule cluster archive

- Naming standards for tapes

- TSM Tape Policy

- Security of dsm.opt and archflt.txt

- Use shared drive or NAS to store scripts

Hock

	<Prev	Next>		<<First	<Prev	Next>	Last>>