Unfortunately, wanting something, and having reality conform to my desires, are two totally different things. So it should come as no surprise that I do not, in fact, have a red Ferrari. Or a Ferrari of any color. Nor is it likely to be the case that I will have one in the future.
And I am OK with that.
When Tony Pearson talks about virtual tape and TSM in his latest post, I get the impression that not only does he want a red Ferrari, he thinks that desire alone is sufficient to ensure that one appears in his driveway.
But lets dissect his post a bit, and see if we can spot the differences between desire and reality. Tony starts off by extolling the virtues of tape over disk in terms of energy efficiency. Without examining those claims here, I will say that I would read the white paper cited with some healthy skepticism. It should come as no surprise that in terms of energy utilization, it is not nearly as black and white as Tony and the authors believe--the game is much closer than they suggest when it comes to the energy utilization of virtual tape over physical tape.
But it is not those comments, but the ones that come after where Tony really runs into trouble.
Tony writes: "But rather than changing over to a VTL, perhaps Mark might be better off investigating the use of IBM Tivoli Storage Manager, using progressive backup techniques, appropriate policies, parameters and settings, to a more energy-efficient IBM tape library. In well tuned backup workloads, the robotics are not very busy. The robot mounts the tape, and then the backup runs for a long time filling up that tape, all the meanwhile the robot is idle waiting for another request."
Really? I don't think so. First, as we have discussed before in A Tale of Two TSMs, the generalization only holds true for the amount of data protected by TSM that is file type, or unstructured data. For everything else, including email and database, TSM behaves like every other backup application. Progressive incremental doesn't matter at all. Second, progressive incremental is very much a double edged sword. It may sound good in theory, but in practice, it is an ugly hog of tape drive resources--due primarily to reclamation activities (and, secondarily, co-location).
The truth is this: in the majority of TSM environments that I have seen, the tape and robots are 80-90% or more utilized. Meaning that of all the backup applications out there, there isn't one that is more demanding of tape drive resources than TSM. Maybe the robot is idle at points, but the way TSM works virtually ensures that you will be driving all available tape drives to a 90% or higher duty cycle.
(The only real way to avoid this is to spend a lot of money on tape drives. Significantly more, in my opinion, than you would require with other backup products to protect the same type and amount of data. Which kind of makes progressive incrementals pointless...)
Then Tony goes on: "If you do decide to go with a Virtual Tape Library, for reasons other than energy consumption, doesn't it make sense to buy it from a vendor that understands tape systems, rather than buying it from one that focuses on disk systems? Tape system vendors like IBM, HP or Sun understand tape workloads as well as related backup and archive software, and can provide better guidance and recommendations based on years of experience. Asking advice about tape systems, including Virtual Tape Libraries, from a disk vendor is like asking for advice on different types of bread from your butcher, or advice about various cuts of meat at the bakery." The emphasis is mine.
At this point, I hope anybody from HP and Sun reading that blog, or this response, is distancing themselves as rapidly as they possibly can from Tony's observations. Because it is quite clear that Tony does not understand, at all, tape workloads in TSM. And if a vendor's understanding of tape workloads is a prerequisite to buying a virtual tape system from them, I think IBM is going to have a lot of hungry sales people.
And, if they don't understand the workload of their own backup application, I despair of them bringing any real understanding to the table when it comes to a discussion of Networker or Netbackup.
Ultimately however, I would invite anybody to come have a talk with us about TSM and virtual tape, and make up your own mind who better understands the challenges, needs, requirements, and use of tape and virtual tape in your infrastructure. I think you will be surprised by the deep, technical, and nuanced conversation that we can have regarding disk and virtual tape in a TSM environment, or most any other backup environment, for that matter.
You probably shouldn't call Tony though. I think he is busy over at the repair shop--his Ferrari just ran into hard reality at full speed, and he did a lot more than just cosmetic damage to that lovely Italian auto.
Scott,
Yet another spot on analysis!
As a large TSM and (temporarily tape...) user this is exactly how it works, I get 90+% utilization on my robot and drives on a daily basis.
How about a compare/contrast of TSM vs. Networker in this area?
Posted by: Craig Frasa | April 18, 2008 at 07:59 AM