From frajaei55 at gmail.com Mon Sep 5 08:24:18 2011 From: frajaei55 at gmail.com (f.rajaei) Date: Mon, 5 Sep 2011 05:24:18 -0700 Subject: [Disksim-users] trace Message-ID: Dear All, I am looking for some ascii traces for SSD, other than postmark and Iozone. Real traces such as OSDB are preferred. A write intensive trace is desired and the bigger is the better (between 200,000 to 500,000 line). Any help would be appreciated. Regards, Farzaneh -------------- next part -------------- An HTML attachment was scrubbed... URL: From haghdoost at gmail.com Mon Sep 5 09:41:14 2011 From: haghdoost at gmail.com (Alireza Haghdoost) Date: Mon, 5 Sep 2011 08:41:14 -0500 Subject: [Disksim-users] trace In-Reply-To: References: Message-ID: You may change the formatting of blktrace traces to ascii. I think Ms. Sameki has written this trace converter. 2011/9/5 f.rajaei > Dear All, > > I am looking for some ascii traces for SSD, other than postmark and Iozone. > Real traces such as OSDB are preferred. A write intensive trace is desired > and the bigger is the better (between 200,000 to 500,000 line). Any help > would be appreciated. > > Regards, > Farzaneh > > _______________________________________________ > Disksim-users mailing list > Disksim-users at ece.cmu.edu > https://sos.ece.cmu.edu/mailman/listinfo/disksim-users > > -------------- next part -------------- An HTML attachment was scrubbed... URL: From vahldiek at mpi-sws.org Thu Sep 8 09:37:04 2011 From: vahldiek at mpi-sws.org (Anjo Vahldiek) Date: Thu, 08 Sep 2011 15:37:04 +0200 Subject: [Disksim-users] Synthetic sequential read performance Message-ID: <4E68C500.608@mpi-sws.org> Dear DiskSim users Currently I'm facing an issue for the clean disksim 4.0 distribution using the Seagate Cheetah model provided on the CMU website in which sequential reads take longer than local reads. I enclosed the used parameter file. (changed the controller to type 3, type 1 shows the same result) I tried to locate the problem using a synthetic seq. read workload and it seems that the following happens: 1. read access (starts at blkno) - seek to cylinder - rotate until first read block - start reading and transfering - finish request 2. read access (starts at blkno + 8) - misses the block blkno + 8 (first block) - reads the following 7 - make one full rotation to read blkno + 8 3. read access (starts at blkno + 16) - rotate to blkno + 16 - read 8 blocks and finish request 4. read access (starts at blkno + 24) - misses block blkno + 24 (first block) - same behaviour as 2. read access Basically every second access introduces a full rotation, because it misses the first block. Unfortunately it does not read during the rotation into a disk buffer segment such that subsequent accesses would return the result immediately. The avg. response time per request is about 4ms which results in a throughput of about 1MB/s. Usually I would expect for a 15k rpm disk 70-80MB/s. Hopefully one of you has seen this before and may be able to tell me how I have to configure or change the disk model parameters. Any idea would be very helpful. Thank you very much in advance. Anjo Vahldiek ----- Max Planck Institute for Software Systems (MPI-SWS) PhD-Student -------------- next part -------------- An embedded and charset-unspecified text was scrubbed... Name: Seagate-Cheetah15k5.parv_base URL: From andy at acunu.com Mon Sep 19 09:55:15 2011 From: andy at acunu.com (Andy Twigg) Date: Mon, 19 Sep 2011 14:55:15 +0100 Subject: [Disksim-users] running a trace without timestamps Message-ID: Does anyone know an easy way to run a given trace of operations (R/W, offset, len) without having explicit timestamps? i.e. the requests enter a queue of a fixed size and the device serves the queued requests as quickly as possible. I'm inclined to modify synthgen to read a trace, but I'm hoping there's an easier way.. Thanks, Andy From leitian.hust at gmail.com Mon Sep 19 12:30:33 2011 From: leitian.hust at gmail.com (Lei Tian) Date: Mon, 19 Sep 2011 11:30:33 -0500 Subject: [Disksim-users] running a trace without timestamps In-Reply-To: References: Message-ID: <312A6862-60E6-4790-85DF-C332CB731467@gmail.com> In general, Once Disksim is starting to handle the current request, it will prepare the next request in the trace. [see the getnextevent() function in disksim.c] Therefore, you may try this way: Assuming the fixed queue length is N, no matter what time the timestamps of the first N requests in the real trace are, you could change the timestamps of them with a very slight ascending order(e.g., 0.000001ms difference) in io_using_next_external_event(). In doing so, DiskSim will enqueue them into its IOdriver request queue and start to process them one by one. Once the threshold N is reached, the No. N+1 request must be given a very large timestamp (e.g., 1000000ms) to make sure this request won't be enqueued and scheduled by Disksim if any one of the last N requests is not done. When one request of the last N requests is completed, you can find the No.N+1 request [its event type is NULL_EVENT] and dequeue it from the intq queue, change its timestamp as simtime + 0.000001ms, and then enqueue it in the intq queue again. In doing so, the length of the request queue of yours is always N, and DiskSim will do the rest for you. Just for your reference. Good Luck! Lei On Sep 19, 2011, at 8:55 AM, Andy Twigg wrote: > Does anyone know an easy way to run a given trace of operations (R/W, > offset, len) without having explicit timestamps? i.e. the requests > enter a queue of a fixed size and the device serves the queued > requests as quickly as possible. > > I'm inclined to modify synthgen to read a trace, but I'm hoping > there's an easier way.. > > Thanks, > Andy > _______________________________________________ > Disksim-users mailing list > Disksim-users at ece.cmu.edu > https://sos.ece.cmu.edu/mailman/listinfo/disksim-users From zh-lian at 163.com Mon Sep 19 21:05:56 2011 From: zh-lian at 163.com (=?GBK?B?wO7V17ui?=) Date: Tue, 20 Sep 2011 09:05:56 +0800 (CST) Subject: [Disksim-users] How to calculate Message-ID: <9eb484.a4e5.132845e2c7e.Coremail.zh-lian@163.com> Hi, Through the output file in the simulation, I found 'Overall I / O System Requests per second: 90.138218', which is too small. How do I calculate the IOPS, throughput and average response time.And how do I define read and write block size. Thanks, Zhaohu Li -------------- next part -------------- An HTML attachment was scrubbed... URL: From liuzhuo9090 at gmail.com Tue Sep 20 18:03:03 2011 From: liuzhuo9090 at gmail.com (liu zhuo) Date: Tue, 20 Sep 2011 17:03:03 -0500 Subject: [Disksim-users] Need up-to-state HDD diskspec Message-ID: Hello, everyone The diskspecs attached within Disksim4.0 seem to be out of date (very small capacity and only for SCSI disks). Is there anyone who has up-to-date HDD diskspec model (*.diskspec, *.model and *.seek files)? For example, SATA disk with hundreds of GB capacity. The dixtrac tool within DiskSim seems only work with SCSI and FC disks. Thanks, On Tue, Sep 20, 2011 at 11:00 AM, wrote: > Send Disksim-users mailing list submissions to > disksim-users at ece.cmu.edu > > To subscribe or unsubscribe via the World Wide Web, visit > https://sos.ece.cmu.edu/mailman/listinfo/disksim-users > or, via email, send a message with subject or body 'help' to > disksim-users-request at ece.cmu.edu > > You can reach the person managing the list at > disksim-users-owner at ece.cmu.edu > > When replying, please edit your Subject line so it is more specific > than "Re: Contents of Disksim-users digest..." > > > Today's Topics: > > 1. Re: running a trace without timestamps (Lei Tian) > 2. How to calculate (???) > > > ---------------------------------------------------------------------- > > Message: 1 > Date: Mon, 19 Sep 2011 11:30:33 -0500 > From: Lei Tian > Subject: Re: [Disksim-users] running a trace without timestamps > To: Andy Twigg > Cc: disksim-users at ece.cmu.edu > Message-ID: <312A6862-60E6-4790-85DF-C332CB731467 at gmail.com> > Content-Type: text/plain; charset=us-ascii > > In general, Once Disksim is starting to handle the current request, it will > prepare the next request in the trace. [see the getnextevent() function in > disksim.c] > > Therefore, you may try this way: > > Assuming the fixed queue length is N, no matter what time the timestamps of > the first N requests in the real trace are, you could change the timestamps > of them with a very slight ascending order(e.g., 0.000001ms difference) in > io_using_next_external_event(). In doing so, DiskSim will enqueue them into > its IOdriver request queue and start to process them one by one. Once the > threshold N is reached, the No. N+1 request must be given a very large > timestamp (e.g., 1000000ms) to make sure this request won't be enqueued and > scheduled by Disksim if any one of the last N requests is not done. When one > request of the last N requests is completed, you can find the No.N+1 request > [its event type is NULL_EVENT] and dequeue it from the intq queue, change > its timestamp as simtime + 0.000001ms, and then enqueue it in the intq queue > again. In doing so, the length of the request queue of yours is always N, > and DiskSim will do the rest for you. > > Just for your reference. Good Luck! > > Lei > > On Sep 19, 2011, at 8:55 AM, Andy Twigg wrote: > > > Does anyone know an easy way to run a given trace of operations (R/W, > > offset, len) without having explicit timestamps? i.e. the requests > > enter a queue of a fixed size and the device serves the queued > > requests as quickly as possible. > > > > I'm inclined to modify synthgen to read a trace, but I'm hoping > > there's an easier way.. > > > > Thanks, > > Andy > > _______________________________________________ > > Disksim-users mailing list > > Disksim-users at ece.cmu.edu > > https://sos.ece.cmu.edu/mailman/listinfo/disksim-users > > > > ------------------------------ > > Message: 2 > Date: Tue, 20 Sep 2011 09:05:56 +0800 (CST) > From: ??? > Subject: [Disksim-users] How to calculate > To: disksim-users at ece.cmu.edu > Message-ID: <9eb484.a4e5.132845e2c7e.Coremail.zh-lian at 163.com> > Content-Type: text/plain; charset="gbk" > > Hi, > > Through the output file in the simulation, I found 'Overall I / O System > Requests per second: 90.138218', > which is too small. How do I calculate the IOPS, throughput and average > response time.And how do I define > read and write block size. > > Thanks, > Zhaohu Li > -------------- next part -------------- > An HTML attachment was scrubbed... > URL: < > http://sos.ece.cmu.edu/pipermail/disksim-users/attachments/20110920/04c339d3/attachment-0001.html > > > > ------------------------------ > > _______________________________________________ > Disksim-users mailing list > Disksim-users at ece.cmu.edu > https://sos.ece.cmu.edu/mailman/listinfo/disksim-users > > > End of Disksim-users Digest, Vol 71, Issue 4 > ******************************************** > -- Zhuo Liu -------------- next part -------------- An HTML attachment was scrubbed... URL: From zh-lian at 163.com Thu Sep 22 20:24:04 2011 From: zh-lian at 163.com (=?GBK?B?wO7V17ui?=) Date: Fri, 23 Sep 2011 08:24:04 +0800 (CST) Subject: [Disksim-users] Need up-to-state HDD diskspec In-Reply-To: References: Message-ID: <696c1444.982.13293aae8da.Coremail.zh-lian@163.com> I have the same problem with you, such as for a large-capacity disk of 80G, dixtrac dose not work. At 2011-09-21 06:03:03,"liu zhuo" wrote: Hello, everyone The diskspecs attached within Disksim4.0 seem to be out of date (very small capacity and only for SCSI disks). Is there anyone who has up-to-date HDD diskspec model (*.diskspec, *.model and *.seek files)? For example, SATA disk with hundreds of GB capacity. The dixtrac tool within DiskSim seems only work with SCSI and FC disks. Thanks, On Tue, Sep 20, 2011 at 11:00 AM, wrote: Send Disksim-users mailing list submissions to disksim-users at ece.cmu.edu To subscribe or unsubscribe via the World Wide Web, visit https://sos.ece.cmu.edu/mailman/listinfo/disksim-users or, via email, send a message with subject or body 'help' to disksim-users-request at ece.cmu.edu You can reach the person managing the list at disksim-users-owner at ece.cmu.edu When replying, please edit your Subject line so it is more specific than "Re: Contents of Disksim-users digest..." Today's Topics: 1. Re: running a trace without timestamps (Lei Tian) 2. How to calculate (???) ---------------------------------------------------------------------- Message: 1 Date: Mon, 19 Sep 2011 11:30:33 -0500 From: Lei Tian Subject: Re: [Disksim-users] running a trace without timestamps To: Andy Twigg Cc:disksim-users at ece.cmu.edu Message-ID: <312A6862-60E6-4790-85DF-C332CB731467 at gmail.com> Content-Type: text/plain; charset=us-ascii In general, Once Disksim is starting to handle the current request, it will prepare the next request in the trace. [see the getnextevent() function in disksim.c] Therefore, you may try this way: Assuming the fixed queue length is N, no matter what time the timestamps of the first N requests in the real trace are, you could change the timestamps of them with a very slight ascending order(e.g., 0.000001ms difference) in io_using_next_external_event(). In doing so, DiskSim will enqueue them into its IOdriver request queue and start to process them one by one. Once the threshold N is reached, the No. N+1 request must be given a very large timestamp (e.g., 1000000ms) to make sure this request won't be enqueued and scheduled by Disksim if any one of the last N requests is not done. When one request of the last N requests is completed, you can find the No.N+1 request [its event type is NULL_EVENT] and dequeue it from the intq queue, change its timestamp as simtime + 0.000001ms, and then enqueue it in the intq queue again. In doing so, the length of the request queue of yours is always N, and DiskSim will do the rest for you. Just for your reference. Good Luck! Lei On Sep 19, 2011, at 8:55 AM, Andy Twigg wrote: > Does anyone know an easy way to run a given trace of operations (R/W, > offset, len) without having explicit timestamps? i.e. the requests > enter a queue of a fixed size and the device serves the queued > requests as quickly as possible. > > I'm inclined to modify synthgen to read a trace, but I'm hoping > there's an easier way.. > > Thanks, > Andy > _______________________________________________ > Disksim-users mailing list >Disksim-users at ece.cmu.edu >https://sos.ece.cmu.edu/mailman/listinfo/disksim-users ------------------------------ Message: 2 Date: Tue, 20 Sep 2011 09:05:56 +0800 (CST) From: ??? Subject: [Disksim-users] How to calculate To:disksim-users at ece.cmu.edu Message-ID: <9eb484.a4e5.132845e2c7e.Coremail.zh-lian at 163.com> Content-Type: text/plain; charset="gbk" Hi, Through the output file in the simulation, I found 'Overall I / O System Requests per second: 90.138218', which is too small. How do I calculate the IOPS, throughput and average response time.And how do I define read and write block size. Thanks, Zhaohu Li -------------- next part -------------- An HTML attachment was scrubbed... URL: ------------------------------ _______________________________________________ Disksim-users mailing list Disksim-users at ece.cmu.edu https://sos.ece.cmu.edu/mailman/listinfo/disksim-users End of Disksim-users Digest, Vol 71, Issue 4 ******************************************** -- Zhuo Liu -------------- next part -------------- An HTML attachment was scrubbed... URL: