I responded offline on the product, but this is basically what I said....

On the server side there is an agent that gets installed.  That agent is the 
part that talks to upstream on the z/os server.    I have brought up replacing 
the product several times and the windows staff go into a panic.  All of the 
backup/tape processing is so automatic on the mainframe side it is foolproof.  
They don't have to mount tapes, keep track of them, etc.  They just need to 
startup their GUI on the windows server and tell it what to do.

Once a week we do full backups on the servers.  Daily we do incremental.  All 
of this backup is done to virtual tape (yep, the same Virtual tape ALL the z/os 
server jobs use).  After the backups complete, we run an Upstream process to 
stack the backups.  We get about 8 server backups on one TS1120 style tapes.  
So we don't ship off a single tape for each server to Iron Mountain.  We keep 
the backups on a regular cycle for 60 days.  Once a month we keep the backups 
(on certain servers) for a year.

So during the normal process if someone needs a file restore, the windows 
admins just go into the GUI on their desktop select the file they want restored 
and restore it.  The windows server talks to upstream, which gets the data from 
the virtual tape.  No real human intervention required.

When we do a Disaster Recover test (we got one next week) the windows server 
admins tell the z/os server admins which servers they need and in what order.  
The z/os admins then run a REGEN job against the stacked tape which rebuilds 
the backups in the upstream catalog.  The windows admins then run their jobs.  
Basically to restore a windows server the restore parms are "//." And they 
select newest to last full option and upstream does the rest.  

In DR, we are doing bare metal restores to unlike servers.  We run IBM blades 
and large servers here and at DR it is either Dell or HP.  

If and when then z/os server goes away, I would just switch to the upstream 
server that runs on distributed platforms.

As far as fast goes, I think it is just a matter of your tape drive speed.  
When we went from 9840A's to TS1120's, I think a large server restore took 
about 20 minutes.

The same process goes for our zFS dataset backups.

Dennis

-----Original Message-----
From: IBM Mainframe Discussion List [mailto:[email protected]] On Behalf Of 
Jim Marshall
Sent: Thursday, March 11, 2010 5:19 AM
To: [email protected]
Subject: Re: FDR/UPSTREAM for Open Systems on zOS

On Wed, 10 Mar 2010 05:18:22 -0800, bob molerio <[email protected]> 
wrote:

>Is anyone using this to backup open systems clients?
>
Looked at all the products a number of years ago and chose IBM TSM for z/OS 
because of various reasons. But my choice may have been abad one now IBM 
has decided to not support the TSM Server on z/OS any more. The 
advantages of z/OS are many for us especially as far as DR goes. My second 
choice was FDR's product though. Back then a number of things bothered me 
about their implementation. 

1. The pricing was on the amount of data to be stored. This causes one in the 
government to have to over buy to make sure you do not suddenly run out. 

2. The storage of files on tapes had to be really thought though because of 
mixed expiration of data. The tape had to be kept around until the last bit of 
data on the tape expired. There was no way to take 2 or more tapes and pull 
all the non-expired data off to make one new one.  This presented a 
challenge. 

3. On had to keep track of what was on what tape in the case of what was 
current. It was not going to as straight forward as the model TSM has which is 
very much akin to HSM for Open Systems. 

I have told IBM people who are running TSM for z/OS are not going to 
suddenly switch it over and host TSM on Windows, etc, because of the 
various operational issues, etc and also losing the notion of leveraging 
existing 
ATLs, VTS's, Automated Operations, skill sets, Automated Scheduling, etc. 
But they contend that they can get DB2 to perform very well on z/OS to 
handle the TSM database therefore it should not be hosted any more on z/OS. 

I will be taking a 2nd look at FDR/Upstream to learn if it is any easier to 
keep 
track of files on all those tapes and also about pricing models. Oh yes, am 
told 
TSM for z/OS will be around stabilized at V5 and supported until around 2013. 
So folks should be making other plans long before they pull the plug. 

jim 

----------------------------------------------------------------------
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

----------------------------------------------------------------------
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