Comments (3)
  • Add a Comment
  • Edit
  • More Actions v
  • Quarantine this Entry

1 seb_ commented Permalink

You are right. The information available about that is rare and well hidden when it actually should be presented very prominently. There are several opinions about the topic anyway and most of them are based on either ignorance, FUD or cost-cutting... often a combination of the three. <div>&nbsp;</div> I personally like this note from the IBM Tape device drivers Installation and User's Guide: <div>&nbsp;</div> "Using a single Fibre Channel host bus adapter (HBA) for concurrent tape and disk operations is generally not recommended. Tape and disk devices require incompatible HBA settings for reliable operation and optimal performance characteristics. Under stress conditions (high I/O rates for either tape, disk, or both) where disk and tape subsystems share a common HBA, stability problems have been observed. These issues are resolved by separating disk and tape I/O streams onto separate HBAs and using SAN <br /> zoning to minimize contention. IBM is focused on assuring server/storage configuration interoperability. We strongly recommend that your overall implementation plan includes provisions for separating disk and tape <br /> workloads." <div>&nbsp;</div> But when I look at the SAN environments in some of my problem cases from the past I have to wonder: "Who reads manuals these days?" :o)

2 anthonyv commented Permalink

Great comment, thanks the extra good information....

3 jdsdavis commented Permalink

I hate that wording. It makes blanket claims with vague justification based on observations from equipment and device drivers from more than 10 years ago. <div>&nbsp;</div> Multipath drivers take care of much of the disk latency issue. <br /> There are no "HBA settings" to be used for tape configuration. <br /> Backup proxy nodes with 99.9% read from disk and 99.9% write to tape going over a single, bidirectional fibre port are not going to experience problems. <br /> "stability problems have been observed" is so very vague, unsubstantiated, and directly refuted by modern experience. Supposedly this was tied to very old tape drivers causing bus resets and FLOGI renegotiations, but that's just anecdotal. <div>&nbsp;</div> I think for people to make claims, there needs to be an actual backing and basis beyond "this is what I learned in the beginning of time, so it must still be true and irrefutable."