Rocket Engineering Corner – The role of tape in mainframe environments
A Conversation with Engineers at Rocket Software
Topic: The role of tape in mainframe environments
Rocket Software Engineering Participants:
Bryan Smith – Vice President, Research and Development and CTO
Joe Devlin – Managing Director, Research and Development
Kevin Shaw – Managing Director, Research and Development
Steve Marak – Senior Manager, Engineering
Ted Ehrlich – Software Engineer
Don Bolton – Manager, Sales Support
Many Rocket customers have a significant investment in tape, particularly for their IBM Mainframe environments. Bryan Smith, VP of R&D and CTO gathered a group of Mainframe savvy Rocket Software engineers to discuss the role of tape and its future in the mainframe space. Here’s what they had to say:
Bryan Smith: Are tapes dead yet? Will they be? Ever?
Don Bolton: No, tapes are not dead and probably won’t be for an indefinite period of time. Applications and subsystems created for the mainframe had tape in mind as a design point. Non-mainframe platforms simply did not invest in designing systems with tape medium in mind; therefore, tape is often less significant in that space than it is in the mainframe world. Major vendors continue to improve tape, making dramatic improvements in the areas of performance and capacity. Today’s tape technology provides speeds that are actually faster than disk for backup applications. So no, I don’t see tape going away anytime soon.
Bryan: Steve, your team deals a lot with tape technologies and products that help customers who are using tape storage – what’s your differing or common view with what Don said?
Steve Marak: It’s some of both. I agree that tape is not going away completely anytime soon and that technology is improving all the time, but we’re seeing that the use for tape has been changing for a long time. Originally tape is how we exchanged data with our business partners. Today nobody does that anymore. We routinely transfer huge files over the network instead of shipping them a tape. Also, we used to see tape as the primary medium for backup and disaster recovery — now tape is not the primary medium – it’s used as a secondary medium. We’re seeing that a lot of the customers on VM and zLinux, if using tape, are using it more as hierarchical storage rather than for traditional backup and DR.
Joe Devlin: What I hear from OMEGAMON customers is the need for support of mainframe tape management systems like CA-1 and RMM. The need for including tape management in what we monitor in OMEGAMON is as strong as ever; however, often we are not managing magnetic tape, we’re managing virtual tape systems.
Bryan: What’s the trend for virtual tape systems?
Ted Ehrlich: Definitely on the upswing. From the OMEGAMON point of view, we’re coming across loads of customers who are getting rid of their physical tape drives and using virtual tape systems entirely.
Bryan: Why are they doing that?
Ted: Speed. Large customers are dealing with large amounts of data and using disk based systems. But they are using magnetic tape for compliance reasons where they’re required to keep that data for up to seven years
Joe: And it’s not just speed – it’s also the price point. They can do this for enough terabytes where it’s not prohibitive.
Steve: Exactly – some are using VTS to do hierarchical migrations. They don’t want to mess with HSM or other products, so they just set VTS policies – after a couple of weeks or whenever, it automatically rolls out from DASD to physical tape and they don’t have to worry about it.
Ted: Yes, a lot of the VTS’ are thought of as cheaper DASD – lower performance, but cheaper DASD.
Steve: We see that almost all the software that lets you use the advanced features of VTSs is written for z/OS. In the VM and zLinux world we have almost none of that capability. So it’s pretty important in these environments that you can just plug in the VTL and that the policies and features work without a lot of interaction from the operating system.
Bryan: I’m surprised that no one has brought up the topic of cost yet. I have to assume that’s the reason why tape technology is still viable – it has some price performance mark that is desirable for companies.
Don: About a year ago we had a bank ask us if we could recover their data because they updated their system with bad data and it was instantly replicated to the second site. They essentially had bad data in two places. They wished that they had a tape backup that would allow them to recover to the point before the bad data was introduced. Because they didn’t have a tape (or offline) copy of the data, they had to do a manual fix and were down for an extended period of time. So companies must weigh the cost of separate data centers vs. copying to tape. If they can afford separate data centers they’ll do it, but in many cases it makes financial sense to keep a tape backup as well.
Kevin Shaw: When talking about the costs of tape, I think customers are always looking at the price differential between disk and tape. Customers are considering energy costs more and more so. Tape is a “green” device in the sense that you can take the media out and keep it on the shelf without the cost of any power. Additionally, the two curves I would think about are latency and cost. If you can trade-off increased latency for decreased cost, you can do it. SSD, for example, and tape have changing price curves.
Bryan: Steve, you’ve spent a lot of time in the z/Linux and z/VM space – what’s unique in the z/Linux and z/VM world for tapes?
Steve: Again, the lack of standard z/OS software support to access advanced library features is the probably the biggest difference. z/OS offers many interfaces and APIs to get at many of those advanced functions. z/Linux and z/VM have almost none of that. The abstraction layer of virtualized environments makes it difficult for both IBM to implement those features and for us to lobby that they are worth implementing.
Don: But on a data protection side, if you have a z/OS system online and attached, we can back up those drives – not at the level of z/VM mini disks or z/Linux file systems, but we back up and recover at the disk drive level itself.
Steve: Well, interestingly, what I’ve been hearing over the last six months or so – and even more so lately, is that many VM and Linux shops are avoiding the dependency on having a z/OS system to do those things. Hearing that surprises me a little, even though I’m a VM bigot, but that appears to be the case.
Bryan: How will IBM’s direction of System z as a cloud solution affect mainframe tape usage moving forward?
Don: We’re in the process of migrating a company using mainframe into the cloud and really it isn’t any different from other migration. The company still needs three LPARs to run its applications during the move. It’s kind of funny that this company is actually moving from an IBM outsourcer to an IBM zcloud – which is actually another mainframe.
Kevin: What I read into cloud direction is that there is a battle brewing between tape and commodity storage. The price of commodity storage is dropping so low so fast that it might get to the point where it makes more sense to keep everything on disk in a private cloud type of environment. Maybe the result is that everyone will just keep more data, but these dropping price points have to play a role eventually.
Bryan: On the open systems side, LTFS integrates tape into the file system transparently – the host server can access files seamlessly, not knowing that they’re on tape rather than disk. Does z/OS have a similar capability?
Steve: All I can say is that we’re not seeing any chatter nor are we having any customer contact regarding LTFS.
Kevin: It could be that the reason LTFS hasn’t been discussed as much (or at all) in the mainframe space is that the capability is often more integrated with the hardware. For example, the TS7740 makes access to physical tape totally transparent – users don’t want to change their tape processes to address LTFS and in the case of the TS7740, transparent file access on tape is already integrated into hardware.
Bryan: OK — to end this topic: for each of you, what’s the oldest tape drive technology that you’ve had your hands on?
Don: The IBM 729 Tape Unit in 1960. I also used paper tape if that counts.
Joe: I used an audio cassette tape drive on a Radio Shack TRS-80
Ted: An IBM 3420 reel-to-reel system. I also used paper tape early on.
Bryan: Paper tape is my story also – DEC paper tape with the MUMPS operating system that I was loading on to a PDP 1184.
Ted: Telephone switches used to produce paper tape.
Steve: On the mainframe side, the 3420.
Kevin: Rocket’s 3420 that we had in the lab over 10 years ago.
Bryan: Great – thanks to all of you. I appreciated your insights. This conversation was certainly interesting and fun for me and I hope it was for you as well. Stay tuned as we will cover additional topics in this type of forum.
To all you readers out there – we would love to hear your comments on this subject. Please feel free to click on the comment button to submit your comment. Thank you.