Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=dx.net.pl header.i=@dx.net.pl header.b="ZuQc0rlx" X-Greylist: delayed 371 seconds by postgrey-1.37 at lindbergh.monkeyblade.net; Wed, 22 Nov 2023 00:36:04 PST Received: from mail.dx.net.pl (unknown [80.68.239.118]) by lindbergh.monkeyblade.net (Postfix) with ESMTPS id D57FE112 for ; Wed, 22 Nov 2023 00:36:04 -0800 (PST) Precedence: bulk X-Mailing-List: fio@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=dx.net.pl; s=dkim; t=1700641791; h=from:from:reply-to:subject:subject:date:date:message-id:message-id: to:to:cc:mime-version:mime-version:content-type:content-type: content-transfer-encoding:content-transfer-encoding; bh=9SAljk5ws8gNEsXqUmHa6gK3/U8WP0FGANsx8px61pQ=; b=ZuQc0rlxjiHsagfBuoE5g55WGRoJblmQJr51ZmVOBffQ9l/J2TcVGpkyeTiwcUV6Z5bS++ Da/eUTI506oNG5paqDEZf6H4XnccoYT5+RIYqrbt1139sIpfEu31RlcYWZo9kHaxBo1TQR xqZwCG+R7ggnVH2snZL02mG1L8JD7eyTuEcQjQdogB36/8ivRE7/JDhR3WdBAaRJiLIEov uYMwbDUtvkfg+NVgdL1kVzmbWiP+4zgeFW5E1xxOOPREvUY3GEjypvxbQotEycGPBlLgVZ l9sYBk77jwJ1e9i/JBVmgUgXFi9xny/jPZtBi51GVZj1250c6iaqfS6zpcICLw== Authentication-Results: mail.dx.net.pl; auth=pass smtp.mailfrom=bartoszek@dx.net.pl Date: Wed, 22 Nov 2023 09:29:51 +0100 From: bartoszek@dx.net.pl To: fio@vger.kernel.org Subject: How to perform an random read test, with normal distribution Message-ID: <4c92fe6a414a472aaec12b04eae3ffba@dx.net.pl> X-Sender: bartoszek@dx.net.pl Content-Type: text/plain; charset=US-ASCII; format=flowed Content-Transfer-Encoding: 7bit X-Spamd-Bar: / X-Spam-Level: * Hi! I Want to perform a test scenario I have a "huge" file in which I want to read out blocks of data randomly with some distribution. For example 20% od blocks in test file will get ~80% od total IO from the engine. Placement of blocks within the file is not important The only requirement is that some blocks are read out more frequently then others, and the distribution of reads is defined by some parameters Test should simulate a VM image reads during VM startup (or runtime) where not all files are used, and some are used more often then others. Is there a way to test such scenario using fio? Best regards, Bartosz Wieczorek