SAS spinup and mode page proposals Questions

Elliott, Robert (Server Storage) Elliott at hp.com
Mon Oct 14 16:10:55 PDT 2002


* From the T10 Reflector (t10 at t10.org), posted by:
* "Elliott, Robert (Server Storage)" <Elliott at hp.com>
*
Thanks for the review.  Comments below...

> -----Original Message-----
> From: Seto, Pak-lung [mailto:pak-lung.seto at intel.com] 
> Sent: Monday, October 14, 2002 1:04 PM
> To: Elliott, Robert (Server Storage); t10 at t10.org
> Cc: Seto, Pak-lung
> Subject: SAS spinup and mode page proposals Questions
> 
> 
> Bob,
> 
>     I have a few questions about the spinup proposal.
> 
> - SPINUP is only used to spin up the target and nothing will 
> be required to spin down the target?  Only START STOP Unit 
> must be used to spin down a device.  

A target can spin-down because:
a) START STOP UNIT with START bit set to zero (Stop)
b) START STOP UNIT with POWER CONDITION set to STANDBY (3h)
c) Power Condition mode page standby timer expiring
d) START STOP UNIT with POWER CONDITION set to FORCE STANDBY (Bh)

> - What is the purpose of keep sending the SPINUP primitive 
> periodically after the device is already spinup?

After initial power on and spin-up, drives can be spun down
and spun-up by software (see previous list).  Without SPINUP
continuously running, software could spin up too many drives 
simultaneously.

> - Initiator can control SPINUP when direct attached to target 
> with the device driver.  How does expander manage the SPINUP 
> sequencing, I am not sure depending on NVROM is acceptable?  
> Should there be some SMP commands that can be issued by the 
> initiator to the expander PHY to tell it to spinup the device?

I couldn't think of a good standard SPINUP control for expanders
at this time.  After experimenting in SAS-1 the industry might 
have some ideas.  The proposal recommends expanders default to
always sending SPINUP out all phys when no vendor-specific controls
are available.

> - Why controlling SATA devices spinup on expander is not 
> included in the proposal?  If SATA devices are not included 
> then the SPINUP primitive is not necessary (or any SMP 
> commands), using the START STOP UNIT SCSI command will be enough.

We cannot add any features to SATA or SATA drives in T10.  The SATA 
WG and T13 are welcome to incorporate a SPINUP-like feature into 
their specifications/standards if they want.

According to SATA 1.0 and errata, SATA drives spinup automatically 
after OOB.  Therefore, an expander in a system supporting SATA 
drives that needs to worry about power sequencing must delay 
running OOB to stagger initial spinup.

Once running, there's no way to prevent software from using
the STANDBY, IDLE, etc. commands to spin down/spinup drives.  That
would require a feature like the SPINUP primitive, an expander
intercepting ATA commands, or a host driver filtering the ATA
commands.

I could also imagine a system where SATA drives could all spinup
at once but SAS drives could not, since the SATA drives had slower 
RPMs and lower peak power consumption.

> - In 10.3, "after power on, if the taret device has 
> previously received a START STOP UNIT command with the START 
> bit set to zero when it receives a START STOP UNIT command 
> with the START bit set to one, spin-up after receiving the 
> next SPINUP."  How is it going to work with SATA devices?

As described above, it doesn't help SATA drives.

> What is the purpose of synchronizing SPINUP with START STOP 
> UNIT command?

As described above, to prevent software from spinning up too
many drives simultaneously.


--
Rob Elliott, elliott at hp.com
Industry Standard Server Storage Advanced Technology
Hewlett-Packard


*
* For T10 Reflector information, send a message with
* 'info t10' (no quotes) in the message body to majordomo at t10.org




More information about the T10 mailing list