xref: /freebsd/sys/dev/isp/Hardware.txt (revision d0b2dbfa0ecf2bbc9709efc5e20baf8e4b44bbbf)
1
2	Hardware that is Known To or Should Work with This Driver
3
4
50. Intro
6
7	This is not an endorsement for hardware vendors (there will be
8	no "where to buy" URLs here with a couple of exception). This
9	is simply a list of things I know work, or should work, plus
10	maybe a couple of notes as to what you should do to make it
11	work. Corrections accepted. Even better would be to send me
12	hardware to I can test it.
13
14	I'll put a rough range of costs in US$ that I know about. No doubt
15	it'll differ from your expectations.
16
171. HBAs
18
19Qlogic	2100, 2102
20	2200, 2202, 2204
21
22	There are various suffices that indicate copper or optical
23	connectors, or 33 vs. 66MHz PCI bus operation. None of these
24	have a software impact.
25
26	Approx cost: 1K$ for a 2200
27
28Qlogic	2300, 2312
29
30	These are the new 2-Gigabit cards. Optical only.
31
32	Approx cost: ??????
33
34
35Antares	P-0033, P-0034, P-0036
36
37	There many other vendors that use the Qlogic 2X00 chipset. Some older
38	2100 boards (not on this list) have a bug in the ROM that causes a
39	failure to download newer firmware that is larger than 0x7fff words.
40
41	Approx cost: 850$ for a P-0036
42
43
44
45	In general, the 2200 class chip is to be preferred.
46
47
482. Hubs
49
50Vixel 1000
51Vixel 2000
52	Of the two, the 1000 (7 ports, vs. 12 ports) has had fewer problems-
53	it's an old workhorse.
54
55
56	Approx cost: 1.5K$ for Vixel 1000, 2.5K$ for 2000
57
58Gadzoox Cappellix 3000
59	Don't forget to use telnet to configure the Cappellix ports
60	to the role you're using them for- otherwise things don't
61	work well at all.
62
63	(cost: I have no idea... certainly less than a switch)
64
653. Switches
66
67Brocade Silkworm II
68Brocade 2400
69(other brocades should be fine)
70
71	Especially with revision 2 or higher f/w, this is now best
72	of breed for fabrics or segmented loop (which Brocade
73	calls "QuickLoop").
74
75	For the Silkworm II, set operating mode to "Tachyon" (mode 3).
76
77	The web interace isn't good- but telnet is what I prefer anyhow.
78
79	You can't connect a Silkworm II and the other Brocades together
80	as E-ports to make a large fabric (at least with the f/w *I*
81	had for the Silkworm II).
82
83	Approx cost of a Brocade 2400 with no GBICs is about 8K$ when
84	I recently checked the US Government SEWP price list- no doubt
85	it'll be a bit more for others. I'd assume around 10K$.
86
87ANCOR SA-8
88
89	This also is a fine switch, but you have to use a browser
90	with working java to manage it- which is a bit of a pain.
91	This also supports fabric and segmented loop.
92
93	These switches don't form E-ports with each other for a larger
94	fabric.
95
96	(cost: no idea)
97
98McData (model unknown)
99
100	I tried one exactly once for 30 minutes. Seemed to work once
101	I added the "register FC4 types" command to the driver.
102
103	(cost: very very expensive, 40K$ plus)
104
1054. Cables/GBICs
106
107	Multimode optical is adequate for Fibre Channel- the same cable is
108	used for Gigabit Ethernet.
109
110	Copper DB-9 and Copper HSS-DC connectors are also fine. Copper &&
111	Optical both are rated to 1.026Gbit- copper is naturally shorter
112	(the longest I've used is a 15meter cable but it's supposed to go
113	longer).
114
115	The reason to use copper instead of optical is that if step on one of
116	the really fat DB-9 cables you can get, it'll survive. Optical usually
117	dies quickly if you step on it.
118
119	Approx cost: I don't know what optical is- you can expect to pay maybe
120	a 100$ for a 3m copper cable.
121
122GBICs-
123
124	I use Finisar copper and IBM Opticals.
125
126	Approx Cost: Copper GBICs are 70$ each. Opticals are twice that or more.
127
128
129Vendor: (this is the one exception I'll make because it turns out to be
130	an incredible pain to find FC copper cabling and GBICs- the source I
131	use for GBICs and copper cables is http://www.scsi-cables.com)
132
133
134Other:
135	There now is apparently a source for little connector boards
136	to connect to bare drives: http://www.cinonic.com.
137
138
1395. Storage JBODs/RAID
140
141JMR 4-Bay
142
143	Rinky-tink, but a solid 4 bay loop only entry model.
144
145	I paid 1000$ for mine- overprice, IMO.
146
147JMR Fortra
148
149	I rather like this box. The blue LEDs are a very nice touch- you
150	can see them very clearly from 50 feet away.
151
152	I paid 2000$ for one used.
153
154Sun A5X00
155
156	Very expensive (in my opinion) but well crafted. Has two SES
157	instances, so you can use the ses driver (and the example
158	code in /usr/share/examples) for power/thermal/slot monitoring.
159
160	Approx Cost: The last I saw for a price list item on this was 22K$
161	for an unpopulated (no disk drive) A5X00.
162
163
164DataDirect E1000 RAID
165
166	Don't connect both SCSI and FC interfaces at the same time- a SCSI
167	reset will cause the DataDirect to think you want to use the SCSI
168	interface and a LIP on the FC interface will cause it to think you
169	want to use the FC interface. Use only one connector at a time so
170	both you and the DataDirect are sure about what you want.
171
172	Cost: I have no idea.
173
174Veritas ServPoint
175
176	This is a software storage virtualization engine that
177	runs on Sparc/Solaris in target mode for frontend
178	and with other FC or SCSI as the backend storage. FreeBSD
179	has been used extensively to test it.
180
181
182	Cost: I have no idea.
183
1846. Disk Drives
185
186	I have used lots of different Seagate and a few IBM drives and
187	typically have had few problems with them. These are the bare
188	drives with 40-pin SCA connectors in back. They go into the JBODs
189	you assemble.
190
191	Seagate does make, but I can no longer find, a little paddleboard
192	single drive connector that goes from DB-9 FC to the 40-pin SCA
193	connector- primarily for you to try and evaluate a single FC drive.
194
195	All FC-AL disk drives are dual ported (i.e., have separte 'A' and
196	'B' ports- which are completely separate loops). This seems to work
197	reasonably enough, but I haven't tested it much. It really depends
198	on the JBOD you put them to carry this dual port to the outside
199	world. The JMR boxes have it. The Sun A5X00 you have to pay for
200	an extra IB card to carry it out.
201
202	Approx Cost: You'll find that FC drives are the same cost if not
203	slightly cheaper than the equivalent Ultra3 SCSI drives.
204
2057. Recommended Configurations
206
207These are recommendations that are biased toward the cautious side. They
208do not represent formal engineering commitments- just suggestions as to
209what I would expect to work.
210
211A. The simpletst form of a connection topology I can suggest for
212a small SAN (i.e., replacement for SCSI JBOD/RAID):
213
214HOST
2152xxx <----------> Single Unit of Storage (JBOD, RAID)
216
217This is called a PL_DA (Private Loop, Direct Attach) topology.
218
219B. The next most simple form of a connection topology I can suggest for
220a medium local SAN (where you do not plan to do dynamic insertion
221and removal of devices while I/Os are active):
222
223HOST
2242xxx <----------> +--------
225                  | Vixel |
226                  | 1000  |
227                  |       +<---> Storage
228                  |       |
229                  |       +<---> Storage
230                  |       |
231                  |       +<---> Storage
232                  --------
233
234This is a Private Loop topology. Remember that this can get very unstable
235if you make it too long. A good practice is to try it in a staged fashion.
236
237It is possible with some units to "daisy chain", e.g.:
238
239HOST
2402xxx <----------> (JBOD, RAID) <--------> (JBOD, RAID)
241
242In practice I have had poor results with these configurations. They *should*
243work fine, but for both the JMR and the Sun A5X00 I tend to get LIP storms
244and so the second unit just isn't seen and the loop isn't stable.
245
246Now, this could simply be my lack of clean, newer, h/w (or, in general,
247a lack of h/w), but I would recommend the use of a hub if you want to
248stay with Private Loop and have more than one FC target.
249
250You should also note this can begin to be the basis for a shared SAN
251solution. For example, the above configuration can be extended to be:
252
253HOST
2542xxx <----------> +--------
255                  | Vixel |
256                  | 1000  |
257                  |       +<---> Storage
258                  |       |
259                  |       +<---> Storage
260                  |       |
261                  |       +<---> Storage
262HOST              |       |
2632xxx <----------> +--------
264
265However, note that there is nothing to mediate locking of devices, and
266it is also conceivable that the reboot of one host can, by causing
267a LIP storm, cause problems with the I/Os from the other host.
268(in other words, this topology hasn't really been made safe yet for
269this driver).
270
271D. You can repeat the topology in #B with a switch that is set to be
272in segmented loop mode. This avoids LIPs propagating where you don't
273want them to- and this makes for a much more reliable, if more expensive,
274SAN.
275
276E. The next level of complexity is a Switched Fabric. The following topology
277is good when you start to begin to get to want more performance. Private
278and Public Arbitrated Loop, while 100MB/s, is a shared medium. Direct
279connections to a switch can run full-duplex at full speed.
280
281HOST
2822xxx <----------> +---------
283                  | Brocade|
284                  | 2400   |
285                  |        +<---> Storage
286                  |        |
287                  |        +<---> Storage
288                  |        |
289                  |        +<---> Storage
290HOST              |        |
2912xxx <----------> +---------
292
293
294I would call this the best configuration available now. It can expand
295substantially if you cascade switches.
296
297There is a hard limit of about 253 devices for each Qlogic HBA- and the
298fabric login policy is simplistic (log them in as you find them). If
299somebody actually runs into a configuration that's larger, let me know
300and I'll work on some tools that would allow you some policy choices
301as to which would be interesting devices to actually connect to.
302
303
304