From sivas at cs.wisc.edu Sat Feb 2 18:26:13 2013 From: sivas at cs.wisc.edu (Sivasubramanian R) Date: Sat, 02 Feb 2013 17:26:13 -0600 Subject: [Disksim-users] assertion spec != 0 in lp_instantiate() [utils.c] failed Message-ID: <510DA095.2060302@cs.wisc.edu> Hi, I'm new to disksim and I'm trying to get a basic configuration set up with disksim 4.0 I'm simulating one disk (maxtor146g) connected to a controller. However when I run the disksim command, I keep hitting the following assertion failure in lp_instantiate() "assertion failed: in lp_instantiate() (util.c:585): spec != 0: no such type Proc" Could anyone help me out with this? Thanks! -Siva From it.heidari at gmail.com Mon Feb 11 04:33:52 2013 From: it.heidari at gmail.com (Naser Heidari) Date: Mon, 11 Feb 2013 13:03:52 +0330 Subject: [Disksim-users] Fwd: simulating RAID-4 with SSD In-Reply-To: References: Message-ID: *Hi all,* * * *for simulating RAID-4 by using SSD, i have changed ssd-iozone.parv file in this way:* instantiate [ ssd0x0 .. ssd0x3 ] as SSD # system topology topology disksim_iodriver driver0 [ disksim_bus bustop [ ############## HBA 0 ############################# disksim_ctlr ctlrHBA0 [ disksim_bus busHBA0 [ ############## Flash-device Raid Controller ############### ssdmodel_ssd ssd0x0 [], ssdmodel_ssd ssd0x1 [], ssdmodel_ssd ssd0x2 [], ssdmodel_ssd ssd0x3 [] ] # end of bus0 ] # end of HBA0 ############ INSERT MORE HBA's ############################## ###### ] # end of bustop ] # end of driver0 (and system topology) *also i have changed logorg block in **ssd-iozone.parv**:* disksim_logorg org0 { Addressing mode = Array, Distribution scheme = Striped, Redundancy scheme = Parity_disk, # vp - added more ssd elements devices = [ ssd0x0 .. ssd0x3 ], Stripe unit = 128, Synch writes for safety = 0, Number of copies = 2, Copy choice on read = 6, RMW vs. reconstruct = 0.5, Parity stripe unit = 128, Parity rotation type = 1, Time stamp interval = 0.000000, Time stamp start time = 60000.000000, Time stamp stop time = 10000000000.000000, Time stamp file name = stamps } # end of logorg org0 spec *Is there any other changes in .parv file or ssd.c, ssd.h and so on ???* *or that's enough for simulating RAID-4 ???* -- Best Regards, Naser Heidari. -------------- next part -------------- An HTML attachment was scrubbed... URL: From sivas at cs.wisc.edu Tue Feb 12 01:02:47 2013 From: sivas at cs.wisc.edu (Sivasubramanian R) Date: Tue, 12 Feb 2013 00:02:47 -0600 Subject: [Disksim-users] Problems with rotational latenxy Message-ID: <5119DB07.7080702@cs.wisc.edu> Hi, I'm trying to simulate a single disk (HDD) using disksim 4.0. * The I/O subsystem architecture is very simple : driver->bus->controller->bus->disk. * I just need to simulate 1ms latency overall from the time the request came (read from trace) to when it reaches the disk. * The input that I'm trying to simulate is synchronous writes (for which I have an idea of how the output will look like). I need to verify that the disksim is configured so testing it with this input. I have also disabled disk block cache for this reason. * Average Rotational latency of the disk is 11ms. * I'm measuring access time for each request and plotting it against sector number. The output looks very similar to what I'm expecting except for one thing : -> the access times for initial few writes is low. Normally, this is fine as the disk starts at sector 0 and input requests are also in increasing order of sector. -> however, due to latency I have introduced in the controller (bulk sector transfer time : 1.0), it should be something like this : at time 0ms, request for sector 0 comes, it writes and now head points to sector 1 (I write only one sector per request) after 1ms, request for sector 1 comes. Now by this time, the disk head should have moved past sector 1 and hence should complete a full rotation before it comes to sector 1. So the first few sector requests should incur high latency. yes? However I do not find this happening. Does the disk start rotating at time 0 (when request is at the highest level i.e read from file) or does it start when the request actually reaches the disk? If its the latter, is there anyway to move the head past a few sector based on the rotational latency? I'm attaching both the expected and actual latency graphs. Thanks a lot! Best Siva -------------- next part -------------- A non-text attachment was scrubbed... Name: Actual.png Type: image/png Size: 467771 bytes Desc: not available URL: -------------- next part -------------- A non-text attachment was scrubbed... Name: expected.png Type: image/png Size: 84949 bytes Desc: not available URL: From abdul.arfan at gmail.com Tue Feb 19 01:14:15 2013 From: abdul.arfan at gmail.com (Abdul Arfan) Date: Tue, 19 Feb 2013 15:14:15 +0900 Subject: [Disksim-users] service time Message-ID: Dear all, I want to ask, how to get total service time information after running disksim? is this information available in disksim output file? -- Abdul Arfan (+82 10 4449 6652) http://arfan86.blogspot.com