toasters June 1997

toasters@lists.teaparty.net

36 participants
28 discussions

Re: Hello...
by sthaug＠nethelp.no 20 Jun '97

20 Jun '97

> I would be very interested to hear from other people using NetApps for > news spools for any advice on optimally configuring the NetApp for > this purpose. > > e.g., no snapshots, increase the number of inodes, no atime update, > disable NFS over TCP, ...? That last paragraph pretty much describes our situation - and it works like a charm! And yes, we have a private 100 Mbps Ethernet (crossover cable) between the Ultrasparc and the NetApp. We keep active, history and outgoing on local disk, and use the NetApp for articles and over- view. The Ultrasparc/Solaris 2.5.1/NetApp system has been *extremely* stable for us. Not a single crash of any kind during several months of service. It just runs. We're happy :-) Steinar Haug, Nethelp consulting, sthaug(a)nethelp.no

3 4

Re: quick usenet poll
by sthaug＠nethelp.no 20 Jun '97

20 Jun '97

> Are people actually putting history files on the NetApp? Ours isn't. Same here. I did what I thought was the obvious thing - put history, active, outgoing and logs on local disks. Use the NetApp for article and overview spool. Steinar Haug, Nethelp consulting, sthaug(a)nethelp.no

1 0

Re: quick usenet poll
by sthaug＠nethelp.no 19 Jun '97

19 Jun '97

> Just curious folks. How big are the history files out there? > The history.pag? 14 days of history here (just after expire has trimmed it): -rw-rw-r-- 1 news news 407992941 Jun 20 00:50 history -rw-rw-r-- 1 news news 84156772 Jun 20 00:50 history.pag history.pag has holes. Real size is only about half of the apparent size given above. Steinar Haug, Nethelp consulting, sthaug(a)nethelp.no

1 0

test - please ignore
by Tom Yates 19 Jun '97

19 Jun '97

please ignore this test. i apologise for the bandwidth loss. Tom Yates - Unix Chap - The Mathworks, Inc. - +1 (508) 647 7561 MAG#65061 DoD#0135 AMA#461546 1024/CFDFDE39 0C E7 46 60 BB 96 87 05 04 BD FB F8 BB 20 C1 8C "Microsoft (tm) is a single-sofa supplier"

1 0

SUBSCRIBE
by Patrick O'Malley, IAC, 617-393-3415 19 Jun '97

19 Jun '97

1 0

Re: NetApp v Suns
by dheiland＠lucent.com 18 Jun '97

18 Jun '97

> From hitz(a)netapp.com Tue Jun 17 22:22 BST 1997 > > It is important for people to understand that LADDIS is a very hard > load, and in real-world environments people will often see much higher > throughput than LADDIS would seem to predict. I recently got a note > from a customer "complaining" that his F230 was running at a sustained > load of over 3300 ops per second, which is over double the LADDIS > "limit" for that system. > > But don't assume that LADDIS results will limit the performance that > you see in a particular real-world environment. I see your point about NFS ops. I would guess then that if this customer was seeing a high percentage usage of the cpu then that would be the indicator that the box was overloaded. And that's not something one can easily extrapolate from the cpu performance of Suns or any other boxes. Estimating the size of filer needed doesn't seem very straightforward then, except by disk sizes. Dave

1 0

Re: NetApp v Suns
by Ketil Kirkerud 18 Jun '97

18 Jun '97

Dave Heiland <dheiland(a)lucent.com> writes: [...] > What kind of loading are people seeing on > their netapps, e.g. 500nfs ops, 2000 nfs ops, any over the limit of the > F540? I've seen 2223 nfsops/sec on _average_ over 6 hours on one of our 540s. It peaked around 6000 :-) (for a few minutes) By the way - This is from our 330 : Wed Jun 18 10:00:05 MET [statd]: 10:00am up 91 days, 18:07 -1750342549 NFS ops I just _love_ the signed 32-bit nfs-op counter - the log looks quite interesting : Tue Jun 3 14:00:05 MET [statd]: 2:00pm up 76 days, 23:27 2145791234 NFS ops Tue Jun 3 15:00:05 MET [statd]: 3:00pm up 77 days, 27 mins, -2147354925 NFS ops -- ---Ketil Kirkerud, Scandinavia Online I synergistically leverage my paradigms

2 1

Advice/suggestions
by Philip Thomas 17 Jun '97

17 Jun '97

Ladies & gentlemen, We use FAServers at our department since Feb. 1994. Since then we have had five units from model 450, F330 to F540. At one point or other our NFS clients were HP, Digital, IBM/RS6000, SGI, Convex, Apollo and even MAC (via Xinet software). Packages range from Cadence to Framemaker to other semiconductor manufacturing related simulators. We have documented uptime of over 99.7%, which includes scheduled downtime. Its truly a lights out operation. Needless to say I am a big fan of the product, to date. Yet, when a sisterly department was looking into NFS fileservers, FAServer was brushed off their short list, which included only SUN and Auspex. Even after two hours of discussion with one of the key IS member I still was in the dark on why FAServer was not even in the running. >From what little I could gather, the only sour note on FAServer was "Network Appliance lack of industry standards"; whatever that means. My guess is that the other prominent players were pointing to the fact that FAServer file system is not "BSD or System5", their OS is not "UNIX" and so on and so forth. When I look at NetApp's brochure, there is no reference to any "standards" or their membership/affiliation to any of the scores of industry consortiums. I tried my best to get something in writing from NetApp stating their commitment on standards. Folks, I know this is absurd. But I had to do it. Anyway, NetApp conveniently ignored my request. So this is you mission, if you chose so... Please let me know if the alleged "lack of industry standards" effected your operations? If it's not against your company policy, it may help, in future, to know (a) what type of industry is this FAServer used? (b) how many FAServer and approximate total disk space? (c) which part of the world are this installed? (d) how important is standards based tools to your company and why you picked FAServer when they seems to be build on proprietary environment? (e) what is your performance and uptime experience with FAServers? (f) how critical is FAServer's uptime for your company/department's operation? (g) installed since when? (h) any experience you can share between FAServer and Auspex/SUN file servers? Thanking you all in anticipation. Philip Thomas Motorola - ACT, M/S M350 2200 W. Broadway M350 Mesa, AZ 85202 thomas(a)act.sps.mot.com rxjs80(a)email.sps.mot.com (602) 655-3678 (602) 655-2285 (fax)

3 2

NetApp v Suns
by Dave Heiland 17 Jun '97

17 Jun '97

We're evaluating an F540 and I'm wondering how many of you compared them against the Suns. Specifically we currently run 2 Ultra 170's plus arrays which both see about 1200-1400 nfs ops/sec average with frequent peaks of 2000 on one. The one box holds common areas, the other builds. I'm considering upgrading these to 2 Ultra 2's with dual 300Mhz chips (for performance and arrays), or one F630 (200GB). >From what I can see of the LADDIS numbers, the NetApps respond faster than the Ultra 2 range and its not until you get to the Ultra 3000 that the response times are comparable, even though the Ultra 2's would sustain higher nfs ops. Is this something people have seen in practice? I'm initially considering one F630 for price reasons and since the one box sees higher nfs/ops but the other box is the one needing the extra storage. The 'limit' of 4500 nfs/ops on the F630 concerns me in that combining the current nfs ops can easily put use near the upper limit of the box. What I don't have a feel for is how many nfs ops the current workload would actually put on a filer. If it responds faster to clients, then I'd actually expect the nfs ops to be even higher, putting me even nearer the limit or above it (although I know its not limited to 4500 nfs/ops, I assume its response would drop significantly). Have people found this to be the case? Two F540's are a possibility, but we seem to be approaching their upper limit and if we go over there isn't room for expandability, whereas we could (eventually) add a second 630 and split the serving again if we choose to go this route. What kind of loading are people seeing on their netapps, e.g. 500nfs ops, 2000 nfs ops, any over the limit of the F540? Dave -- dheiland(a)lucent.com

3 2

Re: Advice/suggestions
by sthaug＠nethelp.no 17 Jun '97

17 Jun '97

> The file service protocols are the obvious critical standards, but > there are many others that we support as well, including standards for > DNS, telnet, inetd, RPC, XDR, portmapper, lockd, NIS, ARP, RMT, rdate, > NDMP, syslog, SMTP, etc. Most (if not all) of these have RFCs > describing them. We support many of these in client-mode only, such as > NIS, DNS, SMTP, etc., because our goal is not to serve these protocols, > but to be a client to them in order to simplify administration. If NetApp claims to support SMTP, it would be nice if you could fix the bug where SMTP commands from a NetApp box are only terminated with LF, not CRLF as RFC 821 explicitly specifies. (I reported this on 12 March this year, and later heard that it was acknowledged as a bug, bug #4268.) Steinar Haug, Nethelp consulting, sthaug(a)nethelp.no

2 1

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

2003

2002

2001

2000

1999

1998

1997

toasters June 1997