iSCSI with Catalyst 3750

[xenmaster.com]

Hi,

I'm building a new iSCSI SAN network with Cisco Catalyst 3750 (stack mode),  EMC storage and Dell servers.  Attached is the plan, a larger PDF  file is located here: http://xenmaster.com/EMCVMwareESX.aspx.   I need help with verifying that there is NO single point of failure and  it is optimized for best throughput. Please focus from the Cisco ASA  5510 firewall backward.  The other switches are controlled by the  DataCenter.

4x 48ports Cisco Catalyst 3750

     2 switches will serve regular IP network (routed traffic)

     2 switches will serve ISCSI and VMware traffic

The stack master will be one of the regular IP network switch

Thank you.

[Gary Ross]

I see 4 issues (mainly document typos) and 1 possible red flag.

1.)  VLAN 20 has a bit mask of 248.  I'm assuming this is meant to be 24.

2.)  VLAN 30 has a bit mask of 248.  I'm assuming this is meant to be 24.

3.)  Based on you addressing scheme, your IP address for the link between the EMC CX3 SPA and the Cisco 3750TS iSCSI switches for VLAN 201 is incorrect.  Rather than it being 10.10.201.202/24, it should be 10.10.201.201/24.

4.)  On the same Cisco 3750TS iSCSI switch, you have VLAN 400 and I think that should be VLAN 210.

With that mentioned, my focus then moves onto your Dell PowerEdge R900 servers.  Is this a 2 node VMware ESX cluster?  If so, and assuming the number of NIC's show is accurate (i.e. 4 NIC's for 2 servers), I see a red flag in this configuration as far as the storage side of this is concerned.

Gary

[xenmaster.com]

Hi Gary,

Thanks for taking the time to give your feedback.  You are right, they are typos and the new plan has been uploaded here http://xenmaster.com/EMCVMwareESX.aspx

A more detailed and updated diagram is here http://XenMaster.com/VMWareEMCDatacenter.aspx

About the server NIC, each server will have 9 NICs, including backup a NIC designated for backup.

NICs:

1 port 0 iSCSI primary path Fabric A    SP A

   port 1 iSCSI secondary path Fabric B   SP A

2 port 0 iSCSI primary path Fabric A    SP B

   port 1 iSCSI secondary path Fabric B  SP B

3 port 0 VMware HA path 0

   port 1 VMware Vmotion path 1

4 port 0 VMWareVmotion path 0

   port 1 VMWare HA path 1

5 Management path 0

6 Management path 1

7 LAN path 0

8 LAN path 1

9 Server Backup path 0

Do you see any red flag there Gary?  I'd love to have your input on the RAID configation too, if you have the time.  Please use the second link since it's the updated version.  Thanks Gary.

[Gary Ross]

The image on the second link isn't very clear like the first one you posted a link to.  Any chance to get a better image to look over?  Thanks for the clarification on the server NICs.  This will help to determine if you'll run into anything.

I can definitely help with the RAID configuration for ESX on an EMC Clariion as we're currently running 4 very large ESX clusters (8 nodes each) with between 9 - 10TB of disk on each cluster.  We're running 2 EMC Clariion CX4-960's.  One at each campus (Primary and DR) with RecoverPoint doing our replication between the CX's using the CX Splitter.  We've gotten the root password for our RPA's so we could tweak the RPA's to push more data than "right out of the box" since our replication network is using 9k jumbo frames and EMC/RecoverPoint "doesn't support jumbo frames" according to EMC.    It does now.

Gary R.

[xenmaster.com]

Hi Gary,

Thanks for your help thus far.  If you click on the link it will expand to a larger 6 page diagram.  The title section is where you'll find the purpose of that page ie. LAN path, SAN path etc.

We're also looking to replicate the 2 EMC SANs but unfortunately I don't have anywhere to reference.  If you have any docs that may help please post.  I've scoured the web for SAN replication but the only diagrams that show up only show just a line.  That doesn't help really because I'd really like to know what devices are making it possible, what kind of WAN links, do we need a multiprotocol router in between, things like that.  Maybe I'll pose that question after this one once I can get the single datacenter up first.  Thanks Gary!

[Gary Ross]

Got it, thanks.  I grabbed the JPG, not the PDF before.  Sorry about that.

I'm more than happy to help with SAN replication.  I've done HUR (Hitachi), SRDF/A (EMC Symmetrix), and RecoverPoint with SANtap

and CX splitters.

I'll look over the PDF tomorrow and will try and respond by the EOB.

Gary R.

[xenmaster.com]

Gary,

Happy Labor Day.  So what do you think of the plan?  Any comments/suggestions?  Thanks.

[xenmaster.com]

Thanks Gary.   Please take your time.