Mark,

A few problems I see with this article:

1) They purchased the tape libraries based on 2:1 compression and the SATA
with zero compression. Virtual tape software will compress the data on disk
at the same rate as tape.

2) Where did they get this disk storage from? The whole 1st year requirement
can be supported by a single midrange controller from HDS or EMC (and
others) - even less when you take compression into account. 48TB per CU does
not represent current technology.

3) Tape needs redundant copies to protect against failure - Disk uses RAID.
They neglected to allow for the redundant copies of data within a tape based
library. My "gut feeling" is that over 30% of tape data consists of
redundant copies to protect against media failure. You eliminate those
copies when you use disk.

4) Many RAID-5 implementations DO NOT require 20% more disk capacity. For
example 7D+P requires 14% more disk, and 15D+P requires 7% more disk.

And BTW I don't agree with 70% usage in a virtual tape disk pool.

A fair comparison should use the same compression on disk, and allow for
elimination of redundant copies. Using the following, I come up with just
raw 93TB of disk in RAID-5 required to replace 163TB Tape Library.

        Current Tape data       163TB
        2:1 Compression  82TB
        70% used space  116TB
        RAID-5 7D+P             133TB
        Less 30% dupes   93TB

Even if you don't clean up redundant copies, 133TB is still a damn site less
than 232TB. A single midrange controller from HDS or EMC would handle 2-3
years of growth in this example in just 1 or 2 racks.

Then there is the cost of maintaining data on tape? Another post mentioned
5PB of tape:

        1) Are they locked into keeping a museum of tape drive technology so
they can read tapes created 5 years ago? What's the maintenance on those?

        2) How does one check the condition of tapes regularly in 5PB
library? Disk drives have S.M.A.R.T and suchlike, tapes have nothing.

        3) Virtual Tape Libraries on disk are self healing on the fly. A
disk drive failure of any magnitude does not cause an outage due to RAID.
The Recovery Time of a failed MVC is anyone's guess, depending on where the
duplicate copy of the data is located.

Unfortunately I don't think this article provides a fair comparison of tape
and SATA costs in a mainframe environment.

Ron




> 
> An interesting article on this subject:
> 
> Tape and Disk Costs - What It really Costs to Power the Equipment
> http://www.clipper.com/research/TCG2007014.pdf
> 
> From the article:
> 
> Key Findings
> 1. SATA disk system has nearly 26 times higher energy costs than tape
> system.
> 2. SATA disk system acquisition costs about 6.5 times the cost of
> automated
> tape system.
> 3. Assuming electrical rates remain same, the cost to acquire, power and
> cool disk systems for five years is almost 8 times the cost to acquire,
> power, and cool automated tape systems.
> 4. The cost to power and cool equipment must be part of the TCO.
> 
> 
> --
> Mark Zelden
> Sr. Software and Systems Architect - z/OS Team Lead
> Zurich North America / Farmers Insurance Group:  G-ITO
> mailto:[EMAIL PROTECTED]
> z/OS and OS390 expert at http://searchDataCenter.com/ateExperts/
> Systems Programming expert at http://expertanswercenter.techtarget.com/
> Mark's MVS Utilities: http://home.flash.net/~mzelden/mvsutil.html
> 

----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO
Search the archives at http://bama.ua.edu/archives/ibm-main.html

Reply via email to