Upgrading Mission Critical Storage - The Plan.

This week I began upgrading the storage in NWC Inc. Since you, our readers, can access portions of NWC Inc. online, I have to take all of the precautions you do when upgrading storage.

As I upgrade the system, I'm writing a workshop about it, offering readers the opportunity to see what our problems were, and how we resolved them.

But you, dear blog reader, get a bonus. Here on the Storage Blog I'm going to "bring you along for the ride" so to speak. You'll get to hear and see it as it happens, and can leave me comments or suggestions.

For today's entry I'll give you "THE PLAN". This is the outline I laid out for accomplishing the goal.

The Plan
NWC Inc., our simulated Mid-Size enterprise has a problem. Our aging NSS NAS (named SpanStor on our network) must be replaced. Over the last several months, it has suffered more downtime than we can afford, and sometimes while it is up we cannot access data on it. This impedes business, and impacts revenues (not to mention interferes with our RealWorld testing) so we are going to replace it.

For our purposes we have chosen Adaptec?s Snap 18000. This product has received glowing reviews from NWC and NWC Inc staff (including yours truly), and we decided that if we must replace our storage, we should follow our own advice.

The project will span more than a week, and includes reconfiguring many systems to access the new storage. To minimize the amount of change being introduced into NWC Inc?s production network, we are opting to masquerade the Snap as the old NSS (as defined below). This is not a ?100%? solution, but it will take us most of the way without having to change shares on production servers. Only some support functions require that we actually change settings to access the new storage.

With all of that said, here is the outline of ?The Plan? to upgrade our storage. We will keep you updated as this plan progresses, and let you know what speed-bumps we encounter along the path to ?brand new storage?.

1.Mount Adaptec Snap-Server 18000 in NWC Inc. racks.
2.Assign IP Address in NWC, Inc. range
3.Configure Snap to be RAID 5 with a hot spare
4.Configure drives to mirror current SpanStor shares.
a.CIFs shares will be exact replica of SpanStor utilizing 1.0 Gigabytes
(currently have 650 Megabytes available on SpanStor)

b.Set aside space for iSCSI for future use Utilizing remaining
400 Megabytes
5.Set up job to bulk-copy from the NSS SpanStor to the Snap.
Configure this job to run each night.
6.Coordinate with Lori, the CIO of NWC Inc. for date to perform
actual replacement.
7.On that date,

a.Perform final copy.
b.Take SpanStor offline
c.Change IP and name of Snap to match the old IP
and Name of SpanStor
d.Bring Snap back online.
e.Reattach drives on all NWC Inc. Machines and test connectivity.

8.Configure Snap Snapshots to replicate data off.
9.Configure tape drive to back up data from the Snaps
10.Observe production environment daily for a week,
and weekly for the remainder of a month for any oddball
applications that malfunction to insure minimal impact from
this change.

Pictures of the machines in question:

NSS NAS that is being replaced. This 13.5U of machine is 640 Gig.

The new Snap 18000. This 2U of machine is 1400 usable Gig.

Next time I'll go over the early speed bumps I hit while starting to install the new server.