|
Home > Archive > Data Storage > December 2005 > SAN ReDesign
You are viewing an archived Text-only version of the thread.
To view this thread in it's original format and/or if you want to reply to
this thread please [click here]
|
|
| Rodrick Brown 2005-11-24, 5:49 pm |
| We built out our enterprise SAN about 2 1/2 years ago, and like every newbie
we did everything wrong, we bought two director class switches (Brocade
12k) and a 9980 from HDS about 40TB and are hemorrhaging ports like no
tomorrow, so i'm pushing for a redesign, I want to switch to a core edge
topology aquiring two new 48k's that will be our core director switches and
about 6 4100's migrating our production systems onto the 4100's and leaving
our staging/developments hosts on the 12k's, right now we have 1 large
fabric spanning across 3 datacenters, I'm also thinking of creating smaller
fabrics for different departments and our large appliction deployments, ie.
We will be adding about 80 AIX servers to our nextwork in 6-8 months, one
approach I was looking at was creating a new fabric for this new application
enviornment and looking at the possiblity of using Brocade Routers.
At a high level I'm looking to acquire
2 x Brocade 48k as core director switches, only devices that will be
connected will be our HDS 9980, 12 - 9940b drives
6 x Brocade 4100's as more departments and applications are added we might
expand this number
[ SITE A]
[SITE B]
(
DWDM)-----------------------------------4GB-------------------------------(DWDM)
// \\
// \\
[9940's] == [ 48k ] == [ 48k ] == [HDS]
[9940's] == [ 48k ] == [ 48k ] == [HDS]
// || || \\
// || || \\
[4100] [12k] [12k] [4100]
[4100] [12k] [12k] [4100]
So for any SAN guru's here, can you let me know if you see any concerns with
a topology like this ?
This is a large network, and operate like a manage hosting / application
service provider, in a little over three years we have grown from about 40
servers to about 400 and we are constantly expanding so I really trying to
get out of the habbit of redesigning this SAN every 2-4 years :-)
--
Rodrick R. Brown
Senior IT Consultant
http://www.rodrickbrown.com
rodrick.brown[<@>]gmail.com
When in 1986 Apple bought a Cray X-MP and announced that they would use it
to design the next Apple Macintosh, Seymour Cray replied, "This is very
interesting because I am using an Apple Macintosh to design the Cray-2
supercomputer."
| |
| Faeandar 2005-11-29, 5:51 pm |
| On Thu, 24 Nov 2005 18:43:21 GMT, "Rodrick Brown"
<rodrick.brown@gmail.com> wrote:
>We built out our enterprise SAN about 2 1/2 years ago, and like every newbie
>we did everything wrong, we bought two director class switches (Brocade
>12k) and a 9980 from HDS about 40TB and are hemorrhaging ports like no
>tomorrow, so i'm pushing for a redesign, I want to switch to a core edge
>topology aquiring two new 48k's that will be our core director switches and
>about 6 4100's migrating our production systems onto the 4100's and leaving
>our staging/developments hosts on the 12k's, right now we have 1 large
>fabric spanning across 3 datacenters, I'm also thinking of creating smaller
>fabrics for different departments and our large appliction deployments, ie.
>We will be adding about 80 AIX servers to our nextwork in 6-8 months, one
>approach I was looking at was creating a new fabric for this new application
>enviornment and looking at the possiblity of using Brocade Routers.
>
>At a high level I'm looking to acquire
>2 x Brocade 48k as core director switches, only devices that will be
>connected will be our HDS 9980, 12 - 9940b drives
>6 x Brocade 4100's as more departments and applications are added we might
>expand this number
>
> [ SITE A]
>[SITE B]
> (
>DWDM)-----------------------------------4GB-------------------------------(DWDM)
> // \\
>// \\
> [9940's] == [ 48k ] == [ 48k ] == [HDS]
>[9940's] == [ 48k ] == [ 48k ] == [HDS]
> // || || \\
>// || || \\
> [4100] [12k] [12k] [4100]
>[4100] [12k] [12k] [4100]
>
>So for any SAN guru's here, can you let me know if you see any concerns with
>a topology like this ?
>
>This is a large network, and operate like a manage hosting / application
>service provider, in a little over three years we have grown from about 40
>servers to about 400 and we are constantly expanding so I really trying to
>get out of the habbit of redesigning this SAN every 2-4 years :-)
It looks to me that you're planning on a single mesh fabric, my advice
would be against that. You don't mention if you're hosts are dual
attached, but I would recommend that they are.
My personal opinion on this is that, with the number of hosts you
mention and expected growth rate, you should be looking to design wide
and not deep. I'm guessing that a single 9980 is not going to suffice
for too much longer either. The fact that you have 400 hosts on it
currently is amazing, I would expect severe performance problems.
~F
| |
|
| This never should have been designed as a single fabric. The products that
you have purchased are all excellent and top of the line. What you really
need is a top notch SAN designer to come and and redesign your topology.
From the sound of you environment, it won't be cheap but it will be a lot
cheaper that buying new SAN switches, service and support and then finding
out that the gear wasn't the problem at all. It was the design! It may patch
the problem, but it is not really addressing it.
Can you provide more detail to the problems that you are having with the
current design?
Regards,
Rick
"Rodrick Brown" <rodrick.brown@gmail.com> wrote in message
news:dRnhf.15168$ek6.11857@news-wrt-01.rdc-nyc.rr.com...
> We built out our enterprise SAN about 2 1/2 years ago, and like every
> newbie we did everything wrong, we bought two director class switches
> (Brocade 12k) and a 9980 from HDS about 40TB and are hemorrhaging ports
> like no tomorrow, so i'm pushing for a redesign, I want to switch to a
> core edge topology aquiring two new 48k's that will be our core director
> switches and about 6 4100's migrating our production systems onto the
> 4100's and leaving our staging/developments hosts on the 12k's, right now
> we have 1 large fabric spanning across 3 datacenters, I'm also thinking of
> creating smaller fabrics for different departments and our large
> appliction deployments, ie. We will be adding about 80 AIX servers to our
> nextwork in 6-8 months, one approach I was looking at was creating a new
> fabric for this new application enviornment and looking at the possiblity
> of using Brocade Routers.
>
> At a high level I'm looking to acquire
> 2 x Brocade 48k as core director switches, only devices that will be
> connected will be our HDS 9980, 12 - 9940b drives
> 6 x Brocade 4100's as more departments and applications are added we might
> expand this number
>
> [ SITE A] [SITE B]
> (
> DWDM)-----------------------------------4GB-------------------------------(DWDM)
> // \\ // \\
> [9940's] == [ 48k ] == [ 48k ] == [HDS] [9940's] == [ 48k ] == [ 48k ] ==
> [HDS]
> // || || \\ // || ||
> \\
> [4100] [12k] [12k] [4100] [4100] [12k] [12k] [4100]
>
> So for any SAN guru's here, can you let me know if you see any concerns
> with a topology like this ?
>
> This is a large network, and operate like a manage hosting / application
> service provider, in a little over three years we have grown from about 40
> servers to about 400 and we are constantly expanding so I really trying to
> get out of the habbit of redesigning this SAN every 2-4 years :-)
>
> --
> Rodrick R. Brown
> Senior IT Consultant
> http://www.rodrickbrown.com
> rodrick.brown[<@>]gmail.com
>
> When in 1986 Apple bought a Cray X-MP and announced that they would use it
> to design the next Apple Macintosh, Seymour Cray replied, "This is very
> interesting because I am using an Apple Macintosh to design the Cray-2
> supercomputer."
>
| |
| Faeandar 2005-12-05, 5:50 pm |
| On Sun, 04 Dec 2005 10:18:19 GMT, "Rick" <r.malo@verizon.net> wrote:
>This never should have been designed as a single fabric. The products that
>you have purchased are all excellent and top of the line. What you really
>need is a top notch SAN designer to come and and redesign your topology.
>From the sound of you environment, it won't be cheap but it will be a lot
>cheaper that buying new SAN switches, service and support and then finding
>out that the gear wasn't the problem at all. It was the design! It may patch
>the problem, but it is not really addressing it.
I would argue that the Brocade 12k is not "top notch", but other than
that it's good equipment.
~F
>
>Can you provide more detail to the problems that you are having with the
>current design?
>
>
>Regards,
>
>Rick
>
>"Rodrick Brown" <rodrick.brown@gmail.com> wrote in message
>news:dRnhf.15168$ek6.11857@news-wrt-01.rdc-nyc.rr.com...
>
| |
| kwalters@pbs.org 2005-12-07, 5:49 pm |
| Rodrick,
Port counts spiraling out of control can also be addressed by utilizing
technology like blades, VMware, and gateways to iSCSI and NFS/CIFS.
Quite often the majority of individual ports are barely utilized.
Consolidating or funneling technology like those mentioned above
provide much improved port utilization and can save you a load of money
in connection costs, not to mention reducing SAN complexity, cable
complexity, and power complexity. Best practice for some time has been
to go with a dual fabric, and LSANS, VSANS, or iFCP can keep remote
facitlities from joining into a single fabric which is also not
recommended. If you have not already, take a look at the Brocade MPR.
They can be combined into an HA pair and provide SAN routing, FC-IP, as
well as iSCSI.
Ken
Rodrick Brown wrote:
> We built out our enterprise SAN about 2 1/2 years ago, and like every newbie
> we did everything wrong, we bought two director class switches (Brocade
> 12k) and a 9980 from HDS about 40TB and are hemorrhaging ports like no
> tomorrow, so i'm pushing for a redesign, I want to switch to a core edge
> topology aquiring two new 48k's that will be our core director switches and
> about 6 4100's migrating our production systems onto the 4100's and leaving
> our staging/developments hosts on the 12k's, right now we have 1 large
> fabric spanning across 3 datacenters, I'm also thinking of creating smaller
> fabrics for different departments and our large appliction deployments, ie.
> We will be adding about 80 AIX servers to our nextwork in 6-8 months, one
> approach I was looking at was creating a new fabric for this new application
> enviornment and looking at the possiblity of using Brocade Routers.
>
> At a high level I'm looking to acquire
> 2 x Brocade 48k as core director switches, only devices that will be
> connected will be our HDS 9980, 12 - 9940b drives
> 6 x Brocade 4100's as more departments and applications are added we might
> expand this number
>
> [ SITE A]
> [SITE B]
> (
> DWDM)-----------------------------------4GB-------------------------------(DWDM)
> // \\
> // \\
> [9940's] == [ 48k ] == [ 48k ] == [HDS]
> [9940's] == [ 48k ] == [ 48k ] == [HDS]
> // || || \\
> // || || \\
> [4100] [12k] [12k] [4100]
> [4100] [12k] [12k] [4100]
>
> So for any SAN guru's here, can you let me know if you see any concerns with
> a topology like this ?
>
> This is a large network, and operate like a manage hosting / application
> service provider, in a little over three years we have grown from about 40
> servers to about 400 and we are constantly expanding so I really trying to
> get out of the habbit of redesigning this SAN every 2-4 years :-)
>
> --
> Rodrick R. Brown
> Senior IT Consultant
> http://www.rodrickbrown.com
> rodrick.brown[<@>]gmail.com
>
> When in 1986 Apple bought a Cray X-MP and announced that they would use it
> to design the next Apple Macintosh, Seymour Cray replied, "This is very
> interesting because I am using an Apple Macintosh to design the Cray-2
> supercomputer."
|
|
|
|
|