By submitting your email address, you agree to receive emails regarding relevant topic offers from TechTarget and its partners. You can withdraw your consent at any time. Contact TechTarget at 275 Grove Street, Newton, MA.
In terms of "feed speed" where is the measurement taken from? Is it from the primary storage to the secondary storage? Does it stop at the write speed on the tape drive? I'm trying to make sense of the optimal performance ratio of 3:1 feed speed to write speed.
"Feed speed" refers to the speed at which data is made available from your disks. This means the maximum "read speed" from the combination of spindles being referenced and the ability of the cache algorithms to do read ahead and provide data to the path. The more spindles the better and the wider the path the better. (This is where 2Gbit Fibre shines.) The idea here is to be able to keep the tape data buffer as full as possible to avoid the "shoeshine" effect as the tape repositions itself to write the next chunk of data. As long as the "feed speed" of your disk subsystem is faster than the "stream speed" of your tape subsystem, then Backups can run at the maximum throughput of your tape subsystem.
This is rarely the case in network-based backup as the network becomes the actual bottleneck. Using a SAN attached tape subsystem with SAN- attached disks provides enough bandwidth to support streaming your data to the tape heads. DLT runs around 5MB pr second per head and LTO is about 15MB per second per head. So say you have a 10-hour backup window to back up 8TB of data. Using LTO on a SAN you could get about 60GB an hour per tape head (depending on compression ratios), (20GB per hour for regular DLT) so we can now calculate how many tape drives we need in a library to meet our backup window. (8192TB/10 hour window = 819GB hour/60GB hour per head = 13 tape heads.) You also need to consider the speed of the robot for how fast it can swap tapes in and out of the drives. The amount of tapes needed in the library depend on the retention schedule for your data.
If you were trying to do this on a network or through a single backup server, then you need to consider Network bandwidth, IP stack efficiency, network utilization, server speed, server memory, capability of backup software to pre-compress data, compression ratio of data types, threading capability of backup software, size of files, and feed speed of disks.
There are great white papers out there from the tape vendors for figuring all this stuff out.
Editor's note: Do you agree with this expert's response? If you have more to share, post it in our Storage Networking discussion forum.
Related Q&A from Christopher Poelker
RAID can allow for better storage performance and higher availability, and there are many different RAID types. Read a comparison of RAID levels, as ...continue reading
SAN expert Chris Poelker discusses how to change the size of a LUN in a Microsoft cluster server environment.continue reading
SAN expert Chris Poelker compares connecting a SAN with wavelength cabling and dark fiber and discusses the pros and cons of each.continue reading
Have a question for an expert?
Please add a title for your question
Get answers from a TechTarget expert on whatever's puzzling you.