Tom, I just want to highlight that these are what we call the corners of the envelope. I Benchmark these products to see where the extremes are, but we know that customer workloads fall somewhere between 100% hits and 100% misses. YMMV, but they give guidance as to what can be achieved by the components and box-wide.
The three important rules to disk tuning are balanced loads, balanced loads, and balanced loads. It's all about looking at the capabilities of your channels and storage to reduce the impact of skew. You don't have to stripe and mirror everything, just get it to where it's not significant. Ron > -----Original Message----- > From: IBM Mainframe Discussion List [mailto:[email protected]] On Behalf Of > Thomas H Puddicombe > Sent: Tuesday, August 03, 2010 7:30 AM > To: [email protected] > Subject: Re: [IBM-MAIN] share mainframe disk experience > > Thanks, Ron. > > Based on your response - and despite some folks assertions to the contrary > - I think I've some I/O "tuning" opportunities. > > Tom > > > > Tom Puddicombe > Mainframe Performance & Capacity Planning > CSC > > 71 Deerfield Rd, Meriden, CT 06450 > ITIS | (860) 428-3252 | [email protected] | www.csc.com > > This is a PRIVATE message. If you are not the intended recipient, please > delete without copying and kindly advise us by e-mail of the mistake in > delivery. > NOTE: Regardless of content, this e-mail shall not operate to bind CSC to > any order or other contract unless pursuant to explicit written agreement > or government initiative expressly permitting the use of e-mail for such > purpose. > > > > From: > Ron Hawkins <[email protected]> > To: > [email protected] > Date: > 08/03/2010 09:17 AM > Subject: > Re: share mainframe disk experience > > > > Tom, > > This was 64x4Gb FICON Channels running from a z9-708. We stopped at 385K > IOPS because we only had 4xSAP and they ran out of puff before the USP-V. > We > got roughly the same result with 32x4GB FICON except response time was > 0.8ms. > > The box had 256GB of cache and 1024 drives installed in RAID-5 7D+1P Array > Groups. > > I always thought 135K SSCH/sec at 100% read misses and 9ms response time > was > a pretty good number also. > > Ron > > > -----Original Message----- > > From: IBM Mainframe Discussion List [mailto:[email protected]] On > Behalf Of > > Thomas H Puddicombe > > Sent: Tuesday, August 03, 2010 5:55 AM > > To: [email protected] > > Subject: Re: [IBM-MAIN] share mainframe disk experience > > > > I agree - 400,000 SSCH/sec @ .7ms is fast. Curious as to the > > configuration of both the channel subsystem and the USP-V. > > > > Tom Puddicombe > > Mainframe Performance & Capacity Planning > > CSC > > > > 71 Deerfield Rd, Meriden, CT 06450 > > ITIS | (860) 428-3252 | [email protected] | www.csc.com > > > > This is a PRIVATE message. If you are not the intended recipient, please > > delete without copying and kindly advise us by e-mail of the mistake in > > delivery. > > NOTE: Regardless of content, this e-mail shall not operate to bind CSC > to > > any order or other contract unless pursuant to explicit written > agreement > > or government initiative expressly permitting the use of e-mail for such > > purpose. > > > > > > > > From: > > Ron Hawkins <[email protected]> > > To: > > [email protected] > > Date: > > 08/03/2010 03:14 AM > > Subject: > > Re: share mainframe disk experience > > > > > > > > Brian, > > > > I'll agree with this to some extent if all you want to do is write a > file, > > read and update it a few times, and then delete it. > > > > However when it comes to Business Continuance, Interoperability, > > Virtualization, and other not so traditional features I don't agree that > > one > > can just go out and by 'NO-NAME' generic mainframe DASD, plug it in and > > expect it to automatically meet all your needs. This where each vendor > has > > strengths and weaknesses that may or may not be relevant to you > companies > > not-so-traditional DASD requirements. > > > > Old Sharks are very fast? I have some difficulty with that one. A USP-V > > doing nearly 400,000 SSCH/sec with 0.7ms response time is what I call > > fast, > > and that's without zHPF. > > > > Ron > > > > > -----Original Message----- > > > From: IBM Mainframe Discussion List [mailto:[email protected]] On > > Behalf Of > > > Brian Westerman > > > Sent: Monday, August 02, 2010 11:23 PM > > > To: [email protected] > > > Subject: Re: [IBM-MAIN] share mainframe disk experience > > > > > > I hate to admit it, but Ted is absolutely right. > > > > > > Most (I want to say all, but just can't commit to it:) vendors units > are > > > very reliable, and even older RVAs, like IBM Shark are very reliable > and > > now > > > are very reasonably priced. > > > > > > Mostly what you get now is increased speed and capacity with a smaller > > > footprint and less energy usage, but even the old Sharks are VERY > fast. > > > > > > Unless you have some mission critical stuff, or applications that have > > > requirements that are far above what the older hardware can handle, > you > > can > > > end up saving upwards of 90% of the cost of a comparable (and faster) > > new > > > unit. If you need the extra features, then by all means, buy it. If > > you > > > don't then why spend money you don't have to? > > > > > > Brian > > > > > > ---------------------------------------------------------------------- > > > For IBM-MAIN subscribe / signoff / archive access instructions, > > > send email to [email protected] with the message: GET IBM-MAIN INFO > > > Search the archives at http://bama.ua.edu/archives/ibm-main.html > > > > ---------------------------------------------------------------------- > > For IBM-MAIN subscribe / signoff / archive access instructions, > > send email to [email protected] with the message: GET IBM-MAIN INFO > > Search the archives at http://bama.ua.edu/archives/ibm-main.html > > > > > > > > ---------------------------------------------------------------------- > > For IBM-MAIN subscribe / signoff / archive access instructions, > > send email to [email protected] with the message: GET IBM-MAIN INFO > > Search the archives at http://bama.ua.edu/archives/ibm-main.html > > ---------------------------------------------------------------------- > For IBM-MAIN subscribe / signoff / archive access instructions, > send email to [email protected] with the message: GET IBM-MAIN INFO > Search the archives at http://bama.ua.edu/archives/ibm-main.html > > > > ---------------------------------------------------------------------- > For IBM-MAIN subscribe / signoff / archive access instructions, > send email to [email protected] with the message: GET IBM-MAIN INFO > Search the archives at http://bama.ua.edu/archives/ibm-main.html ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
