Interview with Fujitsu on SAS Drive Test and Interoperability

Stephen Schaeffer, Manager of University of New Hampshire
InterOperability Laboratory (UNH-IOL) iSCSI, Fibre Channel,
SAS and SATA consortiums, interviewed Mike Fitzpatrick of
Fujitsu Computer Products of America, to learn how the company
approaches testing Serial Attached SCSI (SAS) hard disk drives.
Fitzpatrick was interviewed during a recent SAS interoperability
plugfest held at UNH-IOL that drew 17 companies. Fujitsu and
UNH-IOL are members of the SCSI Trade Association (STA).

Q.
Are there any special challenges to testing dual port drives
as opposed to single port drives? Can you test both dual ports
at the same time?

A.
A dual port drive, such as SAS, has two independent channels
or ports, channel controls and FIFO buffers. Everything behind
the FIFO buffers is shared logic including the cache buffer.
For example, on a read command data is read into the single
cache buffer and then each channel fetches its own data from
the cache buffer. Each port should be tested independently.
The challenge in testing a dual port drive is to test both
ports simultaneously. In testing a dual port drive, one has
to make sure there are no data errors or any errors in managing
the two independent channels.

Q.
In what ways can good testing benefit the cost of storage
to the end user?

A.
The last thing an end user wants to find after installing
a new system is problems. Once an end user has data stored
on a new disk subsystem, any problem that arises is interruptive
and potentially costly. That is why STA and the companies
designing SAS products are putting time and effort into SAS
interoperability testing.

The emphasis
is on interoperability testing long before products are deployed
into the field in order to find and eliminate potential problems
before they occur. Interoperability testing is done to ensure
that vendors are striving to meet the open SAS standard and
that all products will work together.

Q.
What other components in a SAS system with both SAS and SATA
drives should be tested?

A.
One approach to SAS and SATA disk drives is to consider SAS
a superset of SATA. A SAS enclosure allows the user to install
either a SAS or SATA disk drive in the same drive slot. A
SATA enclosure accepts only SATA drives. In a SAS system that
supports both SAS and SATA drives, the SAS Host Bus Adapter
(HBA), the software drivers for the HBA and any expanders
that are between the HBA and the disk drives must all support
both the Serial SCSI Protocol (SSP) and SATA Tunneled Protocol
(STP). The entire system, from the HBA software drivers on
the host side to the drives at the far end, must be tested
against both protocols. SSP is the protocol that talks to
SAS drives while STP is the protocol for SATA drives.

Q.
Have you tested SAS drives that required tester or test method
customization in order to operate within the specification
parameters?

A.
SAS is a new protocol with a new physical level. Any testing
of the physical layer connecting a SAS drive to either a SAS
HBA or SAS tester requires a new level of test equipment and
procedures to ensure the physical layer protocol is operating
properly.

There
are at least four tester/analyzer companies that are in the
process of developing the hardware and test scripts that will
test a product to every aspect of the SAS specification. Prototypes
of these testers are available today.

Q.
Has SAS or SATA software complicated testing SAS/SATA drives?
If so, in what way has it impacted testing?

A.
Except for the fact that new tests had to be developed for
the new physical protocol layer (just like any new interface),
very little if any additional testing has to be conducted
at the disk drive level.

In a SAS
system environment, at the back end of an edge expander port,
or a port on the HBA (if no expanders are involved), there
will be either a SAS or SATA disk drive. It is up to the system
during discovery to determine what type of drive is connected
to the port. It is also up to the system to issue the read
and write commands using the proper protocol. A SAS disk drive
has no knowledge of what a SATA disk drive looks like and
vice versa. The drives act independently of each other.

Q.
What are the market forces driving testing of SAS/SATA disk
drives and what are the priorities?

A.
When it became evident that parallel SCSI would have to be
replaced with a serial technology and work began on SAS, it
took very little effort to convince vendors that the industry
had to conduct interoperability testing before the product
was shipped. Most of the vendors designing SAS-related products
today also ship other I/O products. They understand the need
for testing.

SAS is
the priority because it is the replacement for parallel SCSI,
the predominant interface used today in enterprise disk drives.
Ultra320 SCSI is the last of the parallel SCSI interfaces.
It will inevitably be replaced by SAS drives, thereby requiring
that SAS become stable and ready to be deployed. SATA is a
new interface to the enterprise environment. Projections are
that it will be deployed mostly in new applications such as
near-line storage. At the current time, SATA is not considered
an interface for main line storage like SCSI and Fibre Channel.

Q.
What advice would you give SAS drive vendors about testing
their products?

A.
The issue is not how a SAS disk drive is tested. More
important is how to test a SAS system/subsystem. In the enterprise
environment, the end user will be very certain that any new
interface has done adequate interoperability testing before
it is deployed as a mainstream product. The expectations are
high on the part of the end user and therefore the SAS infrastructure
will have to be tested to a higher standard and proven to
be interoperable with many vendors’ products.

#
# #

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.