-ISCSI over IPoIB (two DDR PCIe 1.0 ConnectX HCA's connected back to back). The results
-for the buffered I/O test with a block size of 512K (initiator)
-against a file of 1GB residing on a tmpfs filesystem on the target are
-as follows:
+Setup:
+
+Target: 2.6.29 kernel, 64 bit, Intel E8400 CPU @ 3.00GHz, 4 GB RAM, SCST trunk
+ revision 727 (which is close to the 1.0.1 release). A file of 1 GB residing
+ on a tmpfs filesystem has been exported via SCST.
+
+Initiator: 2.6.29 kernel, 64 bit, Intel E6750 CPU @ 2.66 GHz, 2 GB RAM,
+ openSUSE 11.0 userspace.
+
+Network: two MHGH28-XTC (MT26418) ConnectX InfiniBand HCA's connected back to
+ back, which are DDR PCIe 1.0 HCA's. The IPoIB stack was configured with the
+ default MTU of 2044 bytes on both interfaces and was using datagram mode.
+ ib_read_bw reported a throughput of 1394 MB/s for this network, and netperf
+ reported a TCP/IP throughput of 1200 MB/s (with default parameters).
+
+Results:
+
+Buffered I/O, block size of 512K (dd if=/dev/sdb of=/dev/null bs=512K):
write-test: iSCSI-SCST 243 MB/s; IET 192 MB/s.
read-test: iSCSI-SCST 291 MB/s; IET 223 MB/s.
-And for a block size of 4 KB:
+Buffered I/O, block size of 4 KB (dd if=/dev/sdb of=/dev/null bs=4K):
write-test: iSCSI-SCST 43 MB/s; IET 42 MB/s.
read-test: iSCSI-SCST 288 MB/s; IET 221 MB/s.
Or: depending on the test scenario, SCST transfers data between 2% and
-30% faster via the iSCSI protocol over this network.
+30% faster via the iSCSI protocol over this network than IET.
+
+Note: at least for the tests with a block size of 4 KB, the initiator
+system was the bottleneck, not the target system.
Something that is not relevant for this comparison, but interesting to
know: with the SRP implementation in SCST the maximal read throughput
</th><th> <a href="http://scst.sourceforge.net/" title="http://scst.sourceforge.net/" rel="nofollow">SCST</a>
</th><th> <a href="http://stgt.berlios.de/" title="http://stgt.berlios.de/" rel="nofollow">STGT</a>
</th><th> <a href="http://iscsitarget.sourceforge.net/" title="http://iscsitarget.sourceforge.net/" rel="nofollow">IET</a>
-</th><th> <a href="http://linux-iscsi.org/" title="http://linux-iscsi.org/" rel="nofollow">LIO v3.0</a>
+</th><th> <a href="http://linux-iscsi.org/" title="http://linux-iscsi.org/" rel="nofollow">LIO</a>
</th></tr>
<tr bgcolor="#E0E0E0">
<sup><A HREF="#6">6</A></sup></td>
</tr>
<tr>
-<th align="left"> Zero-Copy </th> <td> **** </td> <td> ***
- <sup><A HREF="#7">7</A>
- </sup> </td> <td> **** </td> <td> **** </td>
+<th align="left"> Zero-copy passing data between target and
+backend drivers</th> <td> + <sup>
+ <A HREF="#7">7</A>
+ </sup></td> <td> +
+ <sup><A HREF="#8">8</A>
+ </sup> </td> <td> + </td> <td> + </td>
</tr>
<tr>
<th align="left"> Support for transports without expecting
</tr>
<tr>
<th align="left"> Local access to emulated backstorage devices
-<sup><A HREF="#8">8</A></sup></th> <td> + </td> <td> - </td> <td> - </td> <td> - </td>
+<sup><A HREF="#9">9</A></sup></th> <td> + </td> <td> - </td> <td> - </td> <td> - </td>
</tr>
<tr>
<th align="left"> Advanced devices access control
-<sup><A HREF="#9">9</A></sup></th> <td> + </td> <td> - </td> <td> - </td> <td> - </td>
+<sup><A HREF="#10">10</A></sup></th> <td> + </td> <td> - </td> <td> - </td> <td> - </td>
</tr>
<tr>
<th align="left"> Support for Asynchronous Event Notifications
</tr>
<tr>
<th align="left"> Bidirectional Commands </th> <td> + <sup>
- <A HREF="#10">10</A>
+ <A HREF="#11">11</A>
</sup></td> <td> + </td> <td> - </td> <td> - </td>
</tr>
<tr>
<th align="left"> Extended CDB (size >16 bytes) </th> <td> + <sup>
- <A HREF="#10">10</A>
+ <A HREF="#11">11</A>
</sup></td> <td> + </td> <td> - </td> <td> - </td>
</tr>
<tr>
<small> (Windows 2003 clustering) </small></th> <td> + </td> <td> + </td> <td> + </td> <td> + </td>
</tr>
<th align="left"> Safe RESERVE/RELEASE implementation according to
-SCSI requirements <sup><A HREF="#11">11</A></sup></th> <td> Safe </td> <td> Safe </td> <td> Not safe </td> <td> Not safe </td>
+SCSI requirements <sup><A HREF="#12">12</A></sup></th> <td> Safe </td> <td> Safe </td> <td> Not safe </td> <td> Not safe </td>
</tr>
<th align="left"> Safe implementation of Task Management commands
-<sup><A HREF="#12">12</A></sup></th> <td> Safe </td> <td> Not safe </td> <td> Not safe </td> <td> Not safe </td>
+<sup><A HREF="#13">13</A></sup></th> <td> Safe </td> <td> Not safe </td> <td> Not safe </td> <td> Not safe </td>
</tr>
<tr>
<th align="left"> Persistent Reservations <br/>
</tr>
<tr>
<th align="left"> SCSI MIBs </th> <td> - </td> <td> - </td> <td> - </td> <td> -
- <sup><A HREF="#13">13</A></sup></td>
+ <sup><A HREF="#14">14</A></sup></td>
</tr>
<tr>
<th align="left"> Cluster Storage Integration </th> <td> - </td> <td> - </td> <td> - </td> <td> VHACS </td>
<th align="left"> SRP </th> <td> + </td> <td> - </td> <td> - </td> <td> - </td>
</tr>
<tr>
-<th align="left"> iSER </th> <td> - </td> <td> + </td> <td> - </td> <td> - </td>
+<th align="left"> iSER </th> <td> - </td> <td> +
+ <sup><A HREF="#3">3</A>
+ </sup> </td> <td> - </td> <td> - </td>
</tr>
<tr>
<th align="left"> Parallel (Wide) SCSI </th> <td> + </td> <td> - </td> <td> - </td> <td> - </td>
</tr>
<tr>
<th align="left"> SCSI pass-through </th> <td> + </td> <td> - </td> <td> - </td> <td> Single initiator only, not enforced
- <sup><A HREF="#14">14</A></sup></td>
+ <sup><A HREF="#15">15</A></sup></td>
+</tr>
+<tr>
+<th align="left"> Zero-copy data read/write to backstorage</th> <td>BLOCKIO, pass-
+ through <sup>
+ <A HREF="#16">16</A>
+ </sup></td> <td> - </td> <td> BLOCKIO </td> <td> BLOCKIO, pass-
+ through </td>
</tr>
<tr>
<th align="left"> 4k sectors support in pass-through mode </th> <td> + </td> <td> - </td> <td> - </td> <td> ? </td>
Netlink</td> <td> ConfigFS/IOCTL/ProcFS </td>
</tr>
<tr>
+<th align="left"> Zero-copy data send/receive</th> <td> Send only<sup>
+ <A HREF="#17">17</A>
+ </sup> </td> <td> None </td> <td> Send only</td> <td> Send only </td>
+</tr>
+<tr>
<th align="left"> Multiple connections per session (MS/C) </th> <td> - </td> <td> - </td> <td> - </td> <td> + </td>
</tr>
<tr>
</tr>
<tr>
<th align="left"> Bidirectional Commands </th> <td> + <sup>
- <A HREF="#10">10</A>
+ <A HREF="#11">11</A>
</sup></td> <td> + </td> <td> - </td> <td> - </td>
</tr>
<tr>
<th align="left"> Extended CDB (size >16 bytes) </th> <td> + <sup>
- <A HREF="#10">10</A>
+ <A HREF="#11">11</A>
</sup></td> <td> + </td> <td> - </td> <td> - </td>
</tr>
<tr>
(AEN) </th> <td> + </td> <td> - </td> <td> - </td> <td> - </td>
</tr>
<th align="left"> Safe implementation of connections and sessions
-reinstatement <sup><A HREF="#15">15</A></sup></th> <td> Safe </td> <td> Not safe </td> <td> Not safe </td> <td> Not safe </td>
+reinstatement <sup><A HREF="#18">18</A></sup></th> <td> Safe </td> <td> Not safe </td> <td> Not safe </td> <td> Not safe </td>
+</tr>
+<th align="left"> Safe restart <sup><A HREF="#19">19</A></sup></th> <td> Safe </td> <td> ? </td> <td> Not safe </td> <td> ?</td>
</tr>
<tr>
<th align="left"> iSCSI MIBs </th> <td> - </td> <td> - </td> <td> - </td> <td> -
- <sup><A HREF="#13">13</A></sup></td>
+ <sup><A HREF="#14">14</A></sup></td>
</tr>
</table>
<p><A NAME="3"></A> 3. ISER target driver has long known (since Feb 2008) data corruption problem, which localization hasn't been determined yet and might be
in the STGT core. See <a href="http://lists.berlios.de/pipermail/stgt-devel/2008-February/001367.html">
http://lists.berlios.de/pipermail/stgt-devel/2008-February/001367.html</a> and
- <a href="http://lists.wpkg.org/pipermail/stgt/2009-February/002630.html">http://lists.wpkg.org/pipermail/stgt/2009-February/002630.html</a></p>
+ <a href="http://lists.wpkg.org/pipermail/stgt/2009-February/002630.html">http://lists.wpkg.org/pipermail/stgt/2009-February/002630.html</a>.</p>
<p><A NAME="4"></A> 4. The result "in average" is listed. One target can be better somewhere, another one somewhere else. Although manual tuning of target and
system parameters tends the restore the difference listed in the comparison. You can find example measurements <a href="vl_res.txt">here</a>,
Without the kernel patches SCST performance will be at "****+" level, except for the case, when user space backstorage handler used
with iSCSI-SCST iSCSI target driver, where performance will be at "***+" level.</p>
-<p><A NAME="6"></A> 6. There is no performance data for LIO, which allow to make a direct comparison of it with other targets.
+<p><A NAME="6"></A> 6. There is no performance data for LIO, which allow to make a direct comparison with other targets.
The conclusion was made by source code study only. LIO should have performance on the IET level or less,
because of more processing overhead. It might be much less for small block sizes.</p>
-<p><A NAME="7"></A> 7. Some zero-copy functionality isn't available from user space, sometimes fundamentally.
+<p><A NAME="7"></A> 7. In SCST data are always passed in zero-copy manner between target and backend drivers without need for any
+ additional kernel patches, except in case, when local access (scst_local) used with user space backend. </p>
+
+<p><A NAME="8"></A> 8. Some zero-copy functionality isn't available from user space, sometimes fundamentally.
For instance, zero-copy FILEIO with page cache or zero-copy send to a socket. Also STGT can't use splice() for in-kernel
target drivers, because it has memory management in user space. To use splice() with socket-based user space target drivers
STGT would need a deep redesign of internal interactions between target drivers, core and backend handlers.</p>
-<p><A NAME="8"></A> 8. "Local access to emulated backstorage devices" means that you can access emulated by a SCSI target devices
+<p><A NAME="9"></A> 9. Local access to emulated backstorage devices" means that you can access emulated by a SCSI target devices
locally on the target host. For instance, you can mount your ISO image from emulated by the target
- CDROM device on the locally target host.</p>
+ CDROM device locally on the target host.</p>
-<p><A NAME="9"></A> 9. "Advanced devices access control" means that different initiators can see different sets
+<p><A NAME="10"></A> 10. "Advanced devices access control" means that different initiators can see different sets
of devices from the same target. This feature is essential for hardware targets, which don't have ability
to create virtual targets.</p>
-<p><A NAME="10"></A> 10. Not well tested, because at the moment there is no backend using this functionality.</p>
+<p><A NAME="11"></A> 11. Not well tested, because at the moment there is no backend using this functionality.</p>
-<p><A NAME="11"></A> 11. SCSI requires that if an initiator clears reservation held by another initiator, the reservation holder must be notified
+<p><A NAME="12"></A> 12. SCSI requires that if an initiator clears reservation held by another initiator, the reservation holder must be notified
about the reservation clearance. Otherwise, several initiators can at the same time change supposed to be protected by the
reservation data, which can corrupt them. This is what was called
- <a href="http://communities.vmware.com/thread/53797?tstart=0&start=15">"Russian roulette with your data"</a> by someone
- working for VMware on the VMware community forum.</p>
+ <a href="http://communities.vmware.com/thread/53797?tstart=0&start=15">"Russian roulette with your data"</a> on the VMware
+ community forum by someone working for VMware.</p>
-<p><A NAME="12"></A> 12. After a task management command completed and before the corresponding response was sent to the initiator, who sent that task management
+<p><A NAME="13"></A> 13. After a task management command completed and before the corresponding response was sent to the initiator, who sent that task management
command, all the affected SCSI commands must get into a state, where they can't affect following after
the tasks management response commands from this initiator. This is the safe implementation.
The unsafe implementation only marks all the affected
SCSI command, which initiator will send after it received the task management response thinking
that all the aborted commands actually fully aborted. This could lead to a data corruption.</p>
-<p><A NAME="13"></A> 13. The IETF SCSI-MIB (RFC 4455) is not supported by LIO, only proprietary MIBs SBE-SCSI-MIB, SBE-ISCSI-MIB and SBE-IPS-AUTH-MIB are supported.
+<p><A NAME="14"></A> 14. The IETF SCSI-MIB (RFC 4455) is not supported by LIO, only proprietary MIBs SBE-SCSI-MIB, SBE-ISCSI-MIB and SBE-IPS-AUTH-MIB are supported.</p>
-<p><A NAME="14"></A> 14. LIO doesn't emulate all the necessary SCSI host functionality to allow to share a SCSI device
+<p><A NAME="15"></A> 15. LIO doesn't emulate all the necessary SCSI host functionality to allow to share a SCSI device
in pass-through mode to several initiators. It can only pass SCSI commands from initiator to
the SCSI device. This is safe only if there is only one initiator. This limitation
isn't enforced anyhow by LIO. You can find more technical information about that in
<a href="http://www.mail-archive.com/linux-scsi@vger.kernel.org/msg06911.html">
http://www.mail-archive.com/linux-scsi@vger.kernel.org/msg06911.html</a></p>
-<p><A NAME="15"></A> 15. Connections and sessions reinstatement is, basically, a kind of Task Management command, because it implies commands aborting.
+<p><A NAME="16"></A> 16. Proposal how to implement zero-copy FILEIO in SCST you can find in <a href="contributing.html">
+ Contributing</a> page.</p>
+
+<p><A NAME="17"></A> 17. Doesn't need any kernel patch, except in case, when used with user space backend.</p>
+
+<p><A NAME="18"></A> 18. Connections and sessions reinstatement is, basically, a kind of Task Management command, because it implies commands aborting.
For instance, open-iscsi uses it as a less intrusive substistute for target reset in eh_target_reset_handler() callback. So, similarly
- to <A HREF="#7">(7)</A> above, a safe implementation
+ to the safe task management above, a safe implementation of connections and sessions reinstatement
must not accept SCSI commands from new connection/session until all the SCSI commands in
- being reinstated connection/session get into a state, where they can't affect the new commands.</p>
-
-</p>
+ being reinstated connection/session get into a state, where they can't affect new commands.</p>
+
+<p><A NAME="19"></A> 19. "Safe restart" means that after the iSCSI target restart, all the connected initiators will seamlessly restore all existing before
+ the restart connections. "Not safe" means that, most likely, the connected initiators will fail to restore
+ existing connections with some errors.</p>
<p><small><strong>P.S.</strong> You can find a brief background history of this comparison page <a href="comparison_history.html">here</a>.</small></p>
<h3>Some background history of the <a href="comparison.html">Features comparison</a> page.</h3>
-<p>LIO was the first who has had similar comparison page
-http://www.linux-iscsi.org/index.php/Overview_Page. It was and still is very much
+<p>LIO was the first who has had similar comparison
+<script type="text/javascript">
+document.write("<a href=\"ht");
+document.write("tp://www.linux-is");
+document.write("csi.org/index.php/Main_Page#Feature_Overview\">page<\/a>");
+</script>
+<noscript>page http://www.linux-iscsi.org/index.php/Main_Page</noscript>. It was and still is very much
wrong about SCST. I asked that page author, Nicholas Bellinger, to
correct the wrong items (see <a href="http://sourceforge.net/mailarchive/forum.php?thread_name=49AECE0B.5030009%40vlnb.net&forum_name=scst-devel">my
e-mail</a>), but Nicholas Bellinger refused to do it without any explanations. Moreover, he
to the interested people myself. After all our previous discussions and
with his skills and experience it's nearly impossible to believe that Nicholas Bellinger
didn't know that SCST is a lot more generic than LIO and has zero-copy
-in all the same places, where LIO has. Thus, that comparison page looked like rather a deliberate cheating
+in all the same places, where LIO has. Thus, that comparison page is looking like rather a deliberate cheating
attempt. Seems SCST is so much superior over LIO, so Nicholas Bellinger gave up technical discussions and started
-attacking people's perception about SCST, trying to inspire them the
-opposite.</p>
+attacking people's perception about SCST, trying to inspire them the opposite.</p>
-<p>So, for SCST there was no other way, except to setup own,
+<p>So, we had no choice, except to setup own,
correct comparison. This comparison table turned out to be very useful,
so it was extended to cover all the SCSI target areas.</p>
+<h4>Update</h4>
+
+<p>After our comparison page was published, Nicholas Bellinger corrected the most crying items about SCST. Namely:
+he acknowledged that SCST is a generic engine, marked "zero-copy" item for SCST as "Needs kernel patch" and removed "User Interface" item.
+But "zero-copy" mark for SCST is still cheating, because
+SCST needs kernel patch to provide send zero-copy in the iSCSI target driver (and only in it) only if it is working with user space backend. LIO doesn't
+support user space backend, so, if it has "X" here, then SCST must also has unconditional "X" here.</p>
+
+<p>Moreover, new wrong items where introduced:
+
+<ul>
+
+ <li><span>STGT marked as having Fibre Channel target drivers for Emulex, QLogic and LSI cards.</span></li>
+
+ <li><span>STGT marked as having Fibre Channel QLogic target driver in-tree.</span></li>
+
+ <li><span>SCST was marked as "-" in "Additional Header Segment" and "Bidirectional Commands" items.
+ In reality, SCST and iSCSI-SCST support bidirectional commands as well as iSCSI-SCST supports other values in the
+ Additional Header Segment (namely, extended CDB).</span></li>
+
+ </ul>
+
+</p>
+
</div> \r
</div>\r
</div> \r
http://git.etherboot.org / mirror / scst / .git / commitdiff