Date: Thu, 19 Feb 1998 10:52:19 -0800 (PST) From: Simon Shapiro <shimon@simon-shapiro.org> To: Wilko Bulte <wilko@yedi.iaf.nl> Cc: curt@kcwc.com, freebsd-scsi@FreeBSD.ORG Subject: Re: getting oriented with RAID Message-ID: <XFMail.980219105219.shimon@simon-shapiro.org> In-Reply-To: <199802191814.TAA01167@yedi.iaf.nl>
next in thread | previous in thread | raw e-mail | index | archive | help
On 19-Feb-98 Wilko Bulte wrote: > Ach. Ultrix-11 on PDP11/34 with 2x RK05, now that's ancient ;-) Now that is OLD :-) ... > For single stream reads etc cache is pretty useless. But you want WB > cache > anyway to avoid RAID5 writehole pittfalls. So, e.g. data has been updated > on disk but the corresponding parity block never made it to disk when the > power went out. It is worse than that. For a single, streaming read, the cache actually represents an overhed. this is how you get a 2940 to perform ``better'' than a DPT 3334 :-) On the other extreame, where the dataset is huge, extreamly fragmented and totally random, caching losses it. Consider a multi-terrabyte database with a random access to 512byte blocks... For RAID-5 a cache is almost always useful. An HBA cache is generally more useful on the WRITE side, thn on the READ side (which the O/S tnds to have), but Mark is welcome to contradict me here :-) ---------- Sincerely Yours, Simon Shapiro Shimon@Simon-Shapiro.ORG Voice: 503.799.2313 To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-scsi" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?XFMail.980219105219.shimon>