Reply to Message

View discussion in a popup

Replying to:
BHagenSPI
Enthusiast
Enthusiast

New 6.5 deployment - Design Validation please...

We are implementing a brand new vmware 6.5 infrastructure for our enterprise, and I’d appreciate a “sanity check” from the community.

This is a *small* implementation, but not a small company. Downtime could easily cost hundreds of thousands of dollars per hour, and if our electrical generation plant data goes offline the consequences could be far worse. So we want full replication and business continuity. We have our main HQ and a CoLo in different cites, so we’re going with a stack of 4 hosts in each site, replication, and SRM.

We currently run a vmware 5.5 environment with old hosts, little RAM, SAN boxes that are slow and full. We have a total of about 35Tb of data, with a fairly slow growth rate. We’ll end up with about 60 VMs (mostly Windows servers and a couple linux boxes that will probably go away eventually).

We just purchased 8 new hosts, fully HCL compliant (“roll your own”, not ready-nodes…but essentially the same thing for 1/3 the price). 4 boxes will go to each site. Each box has 512Gb RAM and 14.4Tb (raw) storage (ssd cache drive plus 3x 6Tb spinny disks). 4x 1Gb NICs. 2x 10Gb nics.

We also have 2x 10Gb switches coming.

When we're done, we'll literally be downsizing from 4 full racks in our HQ server room to a single rack. The energy savings alone is going to be huge!

So first, a few questions:

  1. Is an external PSC advisable in this small of a deployment? Or will embedded be plenty?
  2. To move VMs from old to new datacenters, is it this “simple”:
    1. Connect the 6.5 datacenter to the 5.5 SAN via iSCSI
    2. Vmotion the VMs to the 6.5 datacenter
  3. To get redundancy at the NIC level with only 6 nics per host, is the “Physial NICs” section, below, acceptable? Or is there a better way?

Next, the design aspect.

Here’s what I’m thinking as a “datacenter” design. Comments?

vCenter

     Datacenter1(HQ)

          Cluster HQ (vsan)

               Host1

               Host2

               Host3

               Host4

     Datacenter2(CoLocataion)

          Cluster CoLo (vsan)

               Host1

               Host2

               Host3

               Host4

Here’s what I’m thinking for Networking. Comments?

Port Groups:

                LAN

                Replication and DMZ

                vSAN and vMotion

                Management

               

Virtual Switches (standard):

                LAN

                Replication and DMZ

                vSAN and vMotion

                Management

Physical NICs

                Vmnic0 - 1gb – Management, physical switch 1

                Vmnic1 - 1gb - Management, physical switch 2

                Vmnic2 - 1gb – Replication and DMZ, physical switch 1

                Vmnic3 - 1gb - Replication and DMZ, physical switch 2

                Vmnic4 - 10gb – vSAN and vMotion, physical switch 1

                Vmnic5 - 10gb - vSAN and vMotion, physical switch 2

VMKernel NICs

                LAN

                Replication and DMZ

                vSAN and vMotion

                Management

I’m looking mostly for comments that address VMware Best Practices; but also any real-world experience that has proven to be acceptable at the highest level of enterprise, 24x7x365 use.

Thanks; I’m looking forward to this conversation!

Reply
0 Kudos