I have a lot of experience with Netbackup and TSM. What Alex has said here
would be a common opinion of someone that fully understood and successfully
implemented all of these products. Veritas would argue that Netbackup is
capable of many more machines, but they do not tell you how they get there.
The reality is they can handle about 10 to 50 machines of the type mentioned
per media server (these would be comparable to a TSM server without a
database, which does not exist as a TSM option or is needed). Netbackup can
direct these from a common master server, but the database (if you could
call it that) on the master really sucks wind when a hundred clients get
going that have lots of files. With a Windows Master/Media server
environment they simply cannot scale, do not even try it if you plan on
implementing more than 50 total servers (couple terabytes). Duplication for
offsite movement is really poor with Netbackup. In fact you cannot have
more than one duplicate copy. The performance is really horrible if you use
the multiplexing function. If you do not use the multiplexing function then
you have to buy so much tape hardware it is pathetic. Basically, they
recommend tape drives be installed on every server and running the media
server code on each machine. The cost of this media server code is half the
cost of a Master license if the system just saves itself, the same cost if
it saves other servers.
I am guessing in Netbackup's early design the concept of a TSM type master
server was never envisioned really. The issue was people needed a way to
manage all the server backups each having their own dedicated tape hardware
from a central tool. Netbackup does that well up to a point. This is what
sells customers on the Netbackup product. In the days when servers had 30
GB saving the whole system weekly to send it offsite and because the data
was not considered critical sending the backup from the week before offsite
were acceptable implementations. 30GB could be saved in about 3 hours.
Now, these 200 to 500GB servers with 1+M files hit and Netbackup has no hope
of a full save for offsite, no way to consolidate incrementals for offsite
disaster recovery restores, no way to duplicate for offsite storage.
Enter TSM (actually before Netbackup). TSM is not for the meek. It is
truly an enterprise class product. It is a storage management tool that
does backup and restore functions. It takes advantage of its storage asset
management using differential backup technologies and dramatically reducing
the hardware requirements compared to classical full weekly backups with
incrementals during the week. I would like to emphasize differential backup
technologies. Subfile backup as an example. Never more than one copy of a
version of a file unless you force it to do it.
Now, Database servers are a different story. Here the numbers of files
become a non-issue. Until SAN managed tape was available and the TSM SAN
Managed Client you had to put the TSM server on every large machine so that
you could attach direct tape. Well, that created somewhat of a more complex
environment, but TSM had some capabilities to manage all the servers from a
central server. The SAN Managed Client has put Veritas in a predicament in
recent sales opportunities. TSM is now drastically cheaper in these large
environments to operate because of the reduced hardware in comparison.
Remember though what TSM was originally created to do, save PC desktops and
manage hundreds/thousands of them. So, from the ground up scalability was
never an issue. The problem was they were slow to recognize and deliver the
server agents/clients because IBM did not acknowledge applications were a
distributed thing. Now, application support is there for the most part.
Tivoli has a few things to deliver to be whole.
For core enterprise customers no matter whether they are HPUX, Windows, AIX,
or Solaris only, TSM should be the product of choice. Why? Price,
Functionality, Scalability, and understanding of Storage Management. Many
of IBMs mainframe storage management people have moved to Tivoli in San Jose
to support and develop TSM. Why, they have basically finished the job in
the mainframe world and are taking all of this excellent technology,
capability, and something more important, years of experience and lessons
learned, in the behemoth mainframe world and delivering a superior solution
for the Open Systems world.
Yes, I did not answer the question of TSM vs Backup Exec. Again, it is like
comparing a push scooter to an 18 wheeler. There is little in common other
than both have some axles and wheels.
Alex puts it in perspective pretty well.
|
