Unsolved
This post is more than 5 years old
9 Posts
0
1180
June 22nd, 2010 10:00
DMX & VMax FA to SAN Connections.
I'm looking for best practices for FA to SAN connectivity. Right now I have a DMX in a dual SAN switch environment (SAN1 & SAN2). Since this environment has been configured for some time, this question is more "was this done properly?"
7a (A & B) to SAN1
7b (A & B) to SAN2
7c (A & B) to SAN1
7d (A & B) to SAN2
(and this repeats for FAs 8, 9, & 10).
Would it have been better this way:
7aA SAN1 8aA SAN2
7aB SAN2 8aB SAN1
7bA SAN1 8bA SAN2
7bB SAN2 8bB SAN1
7cA SAN1 8cA SAN2
7cB SAN2 8cB SAN1
7dA SAN1 8dA SAN2
7dB SAN2 8dB SAN1
(and repeat this for FAs 9 & 10. and use Rule of 17).
Now for the future...we're looking at VMax (2 engine) what is the best practice for FA to SAN connectivity, and does the Rule of 17 still apply?
Thanks...
David


dynamox
11 Legend
•
20.4K Posts
•
87.4K Points
0
June 22nd, 2010 13:00
i don't believe rule 17 appies to Vmax. As far as zonning is concerned we have been following this practice documented on page 8 and 9
https://powerlink.emc.com/nsepn/webapps/btg548664833igtcuup4826/km/live1/en_US/Offering_Technical/White_Paper/h6209-symmetrix-vmax-vmware-virtual-infrastructure-wp.pdf
JasonBailey
147 Posts
0
June 22nd, 2010 17:00
yeah, rule of 17 doesnt apply anymore
SKT2
2 Intern
•
1.3K Posts
0
June 23rd, 2010 03:00
In VMAX the idea is to distribute the directors across the engines. So even w/o rule 17 u can achieve that. But rule of 17 still make sure that the directors ARE not distributed on the same Engine.
Quincy561
1.3K Posts
0
June 23rd, 2010 04:00
The idea of the rule of 17 was to get redundant connections on different boards that were physically far away from one another.
You still want to do the same thing on VMAX.
And you can put the alternate paths on the same engine, as an engine has two redundant directors. It is better to spread across different engine enclosures if possible. You should not put the redundant paths on the same engine or worse, on the same IO module.
RobertDudley
2 Intern
•
448 Posts
0
June 23rd, 2010 05:00
Rule of 17 was in place when the Symmetrix was a bus architecture. By following rule of 17 you never had host connections on the same bus within the Symmetrix. When the matrix was introduced (DMX3/4 series) it nullifyed the need for the rule of 17 and that is carried on with the V-Max engine build. It tends to be easier if you set a standard for SAN connectivity. With our arrays on the Symmetrix side we place odd directors in our odd san and all even directors to the even san. Even on the V-max we still set our hosts up so that we follow the "rule of 17", its an easy standard to follow.
dynamox
11 Legend
•
20.4K Posts
•
87.4K Points
0
June 23rd, 2010 08:00
i thought rule 17 was all about connecting hosts to resources in different power zones. So on your Vmax you place odd FAs on one fabric and even on another ? I have one engine right now so i put odd and even on each fabric and each HBA gets zoned to odd and even FAs for a total of 4 paths per host (Clariion style zoning).
dtaraian
9 Posts
0
June 23rd, 2010 09:00
Thanks for the help...this thread has cleared up many issues.
In response to RobertDudley's post; putting even directors in one SAN and odd in another, is there any advantage to placing alternating ports in alternating SANs? In other words 8e0, 8f0, 8g0, & 8h0 in SAN1 and 8e1, 8f1, 8g1, & 8h1 in SAN2?
And lastly, when our 2 engine VMax is delivered does the engine numbering start at 1 & 2, or 3 & 4? And what are the actual engine to director mappings; i.e. engine 3 (dirs 5 & 6), engine 4 (dirs 7 & 8)?
RobertDudley
2 Intern
•
448 Posts
0
June 23rd, 2010 11:00
Rule of 17 was in use back in the Symm 3 days where there was an odd and even bus in each array. There were events that could drop the bus (IE termination failure) so as long as the host was on both buses you (ideally) were not impacted. Power zones were brought in with the DMX2000/3000 series and were not a factor in rule of 17.