|
by Patrick Fitzsimmons, Innovation
Data Processing
Here we are coming to the end of this millennium and network and LAN administrators
are doing what they have been doing since the advent of LANS were installed:
still searching for that perfect backup method. As the Wizard of Oz or
was it the Good Witch of the North said...look no further than in your
own backyard...so if you have a mainframe, then look no further! You already
have the perfect backup server!
It seems that shops with distributed data and a MVS mainframe can leverage
their existing infrastructure, hardware, and personnel to manage this
data. Is there really anything different about distributed data that can't
be managed by the Operations Department, NO of course not. Is a megabyte
not a megabyte?
By using the existing resources, such as your automated MVS job scheduler,
very mature robust tape management system, security system you can get
the benefit of getting backups performed in a lights out, automated and
unattended environment. What could be better than that? Ok, in a perfect
world 1 day work weeks and virtual backups would be it.
With ATLs, Robots, and Silos already on the data center floor why not
take advantage of these devices and the available MVS cycles? The mainframe
tape devices have approached and finally reached the point where their
capacities are equal to DASD devices in modern day. The Magstars with
fast access time and 60GB compressed capacity and the STK 9840 also with
60GB compressed capacity finally rival what LAN administrators see on
LAN attached devices. But with an important difference...with much better
reliability.
The management of this LAN/UNIX data can be performed by the MVS tape
management system such as CA-1 or IBM's RMM, using GDGs or retention periods
to manage expirations. These excellent systems have been the standards
in operations of the MVS staff and have been performing adequately for
many years, with nothing coming close to competing with their reliability
in the distributed world.
Many improvements have also come over the last 10 years in network capacity
and bandwidth. When coax was king, so were 286 processors, but now generations
later with Pentium class machines and RISC processors, that combine with
100mbit networks, resulting in much improved access to the mainframe.
Gone with coax and 286s are the older 3174 cluster controller or 3745
Front End processor, being replaced by the OSA adapter and channel connections
that make bulk data transfer orders of magnitude faster allowing the mainframe
to be used as the big backup server. Newer technologies such as ATM, ESCON,
FDDI and FICON are making bulk data transfer a goal and the norm rather
than the exception and something to be avoided.
With 30+ years of protecting mainframe data, the venerable mainframe disaster
recovery plans have been proven and well tested. The compelling question
then is "Why try and duplicate the staffing and hardware" when they can
be used for other more productive and revenue generating goals that aren't
repetitive and are more useful.
So with all this hardware and personnel to manage it why NOT use the mainframe
to backup the distributed data. Innovation Data Processing a leader in
storage management since 1972 has a solution with its FDR/Upstream product.
FDR/Upstream is a storage management product for backup/restores, and
management of PC/LAN/UNIX data. The backups can be automatically initiated
by existing MVS job scheduler by submitting "MVS batch jobs" and the backups
can be written either directly to mainframe tape or DASD. The backup process
can use either TCP/IP or SNA APPC LU6.2 protocols. The "batch job" can
also optionally wait until the completion of the backup job and perform
JCL condition code checking allowing exception handling to take place!
Zero condition code goes to another step in the job or another job, a
non zero condition code can cause pre-programmed notification tasks such
as to beep or page someone.
A most important and desirable benefit is centrally managed backups that
are automatically initiated by the MVS ops group. As for restores they
can be performed by the MVS ops group or can be performed by a few different
departments (making the NT LAN or UNIX group happy). A JAVA or GUI interface
allows the restores to be monitored and performed by either of your colleagues
in the LAN/UNIX group, and allows them to feel like masters of their own
destinies, and avoiding the lively discussion of whose data is more valuable
to the organization.
In Disaster recovery planning FDR/Upstream's vaulting facility can be
used to create a secondary tape copy for DR to be stored in an offsite
vault, and can be used for onsite recovery if primary backups are damaged
or unusable, or for long term recovery.
FDR/Upstream has several data reduction techniques that minimize how much
data is sent across the wire. With Upstream's "Full Merge Backup", you
can obtain a complete full backup, while only sending a fraction of the
total data across the network. The Upstream Full Merge Backup facility
drastically reduces the elapsed time of full backups, by utilizing already
existing mainframe backups of files which have not changed, instead of
transmitting them from the PC.
Using a sophisticated technique, FDR/UPSTREAM can construct a complete
full backup without the PC having to read or send most of the files, resulting
in extraordinary performance for full backups.
To use the full merge backup, a first-time baseline full backup of the
file server is performed, backing up ALL and transmitting all the specified
files to the mainframe. Once it has this baseline, future backups need
only be incremental and full merges. When performing Full Merge Backups,
Upstream sends a picture of the drive to be backed up to the Upstream
MVS database, which uses earlier existing backups plus changed files to
construct the full merges. Because relatively few of the files must be
read or transmitted, the benefits are greatly reduced network traffic
and complexity, and dramatic time savings for full merge backups.
Merge backups include a duplicate file transmission suppression facility.
Often similar files are duplicated on many machines across your enterprise.
These include operating system files, word processors, and other software
packages. FDR/UPSTREAM can be set up to recognize certain files as duplicates
and not transmit them saving transmission time, and reducing host storage
requirements. Determination of duplicate files can be performed manually,
by manually backing up the files which are known to be duplicates to the
duplicate file database. You can also activate the unique Automated method
which will identify duplicate files for you automatically. These features
combined with compression make efficient use of the network bandwidth.
While the average hard drive left megabytes in the rear view mirror for
the now standard gigabyte drives, some shops think they have too much
data to consider a mainframe solution. For these environments, there are
alternatives.
Upstream can be used to backup data across the network or across the mainframe
ESCON channel. Innovation has teamed with EMC to produce software that
takes advantage and exploits the channel connection of the EMC Symmetrix;
resulting in FDRSOS and Upstream/SOS. FDRSOS performs physical volume
backups of the open systems data on the Symmetrix and Upstream/SOS provides
incremental backup processing, file level granularity, and single file
restore capability. Both FDRSOS and Upstream/SOS will also work with the
EMC BCV (Business Continuance Volumes) volumes that make a local, non
disruptive, point-in-time copy of a volume of production data, allowing
the backup to be taken while the production volumes are up and still available
for production processing.
FDR/Upstream can also be used with MVS channel cards. These cards are
installed in the file server, are connected via fiber, use TCP/IP as the
communication protocol, and will then transfer the data across the host
channel at much faster speeds than even the fastest network. Upstream
works with the Polaris ESCON card from Polaris Communications, as well
as other cards such as IBM's Netfinity card. These cards are fast, reliable
and easy to implement mainframe connections from Windows NT/2000 and UNIX
systems. The Polaris 6950 PCI to ESCON Channel Interface directly connects
an open system PCI bus to a mainframe ESCON channel. With onboard RISC
processor and memory to handle the ESCON protocol, the Polaris 6950 can
transfer data bi-directionally using the full bandwidth of IBM's Enterprise
System CONnection (ESCON) fiber based protocol.
Even for large databases it is a good automated solution. FDR/Upstream,
due to its command line control and program return code facilities, can
be integrated into the facilities provided by most database vendors providing
online backup support for Oracle, UDB Universal Server, Lotus Notes, MS
Exchange, SQL Server and is SAP certified. Upstream can also be used with
BMC SQL Backtrack to manage backups for Sybase and Informix databases.
By using the database vendor APIs, Upstream is able to backup the database
online and also an incremental backup, and thereby not requiring a backup
of the complete database everyday, resulting in a much faster backup!
Moreover, if you really need a reason to make the mainframe folks give
their approval, Upstream can also be used to backup/restore OS/390 Unix
System Services, also known as HFS, on a file basis and also has file
transfer facilities with ASCII to EBCDIC translation.
Protecting the corporate data that is distributed on PC LANS and UNIX
servers should not require a complete and separate solution from the mainframe
datacenter. In many companies today the servers are already stored IN
the datacenter, so using the hardware there is a natural extension.
So if you haven't considered backing up to your mainframe in the past,
maybe its time to give it some consideration or a look. What with the
new hardware technologies and Innovation Data Processing's software assistance,
maybe you can "look no further than your own back yard".
|
