One other consideration is if you are encrypting the data or not. Encryption will also slow down the read/write of data.
If you are encrypting, will it be at the hardware level, the tape drive does the encryption, or at the software level, you have a product like EKM/TLKM. I found that in an ATL (3494) it takes about 30 secs to find tape, move tape to drive, drive read header of tape. After that could be your read/write time. Of course, if you have operators mounting your tapes, even longer. ;-) Also, the larger the tape, the longer the access to the back of the tape. If you are looking for a faster tape access process, I would probably also research "tapeless" solutions. With "Tapeless" you do not need to worry about tape speed (it is now disk speed) and you do not need to worry about copying from old media to new media. You can also replicate the "tape" to another "tapeless" solution. No transport of tape and potential loss. Down side - your "tapeless" device can keep growing as your tape library holds more data. That could require more power usage, more cooling requirements, and that could be more expensive than sticking tapes on a shelf. Lizette > -----Original Message----- > From: IBM Mainframe Discussion List [mailto:[email protected]] On > Behalf Of R.S. > Sent: Tuesday, October 29, 2013 5:07 AM > To: [email protected] > Subject: Re: Tape drive speed > > W dniu 2013-10-29 12:54, גדי בן אבי pisze: > > Hi, > > > > I was asked to estimate the change in the time it would take to backup > > disks when > moving from one type of tape drive and cartridge to another. > > We currently have 3590H’s and it takes about 30 minutes to back up a > > 3390-54. > > What would happen if we convert to TS1120’s TS1130’s or TS1140’s. > > Our tape drives are connected using ESCON. > > What would happen if we convert to FICON? > > > > I tried looking at the datasheets, but there are many speed ratings. > > > > We are running z/OS 1.13 on z114, and the backup is done using native > DFSMSdss. > IT DEPENDS. > First, you have to identify the slowest element, bottleneck. In case of FICON- > attached TS1140 it could be your DASD or ...data. > Yes, the numbers presented by vendors assume optimal compression ratio. > Data less compressibel and more compressible give worse performance. Of course > while you can (in theory at least) change your DASD and channels, you cannot > change your data to achieve optimal compression ratio. > BTW: don't forget about mount time: TS1140 has much larger capacity, so much > less cart will be needed, so much less mount/dismount operations will be > needed. > Evey mount is approx. 1 minute, I'm not sure about dismount, but should be > similar. > > My €0.02 > > -- > Radoslaw Skorupka > Lodz, Poland > > > > > ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN
