After months of planning and waiting (and years of grant writing and more planning and waiting), my dev team, blitZEN (and our constituent research centers at the UNC School of Medicine: CCBC and PaTRC [no link yet!]) have now procured, (mostly) configured and racked the first wave of our enterprise hardware stack.

In The Rack

blitZEN Stack: In The Rack

Ain’t it purdy? Check out this gratuitous drive array shot:

the beginning

a-lot-o-storage: the beginning

OOOH That’s niiiice…. OK now, sweetie, why don’t you do a pivot-turn like the super model you are and show them the rest of you:
blitZEN Rack - the flip side

I need to prod the responsible folks to discuss all of the gory details (I didn’t have any input in engineering this but boy I had lots of input-ing them in the damn rack! ;) ), but here’s what we’ve got hardware-wise:

I will verify/update this for accuracy. This is off the cuff, strictly from memory

  • 4 1-U Silicon Mechanics servers. Dual Quad-core (Intel processors I believe). Right now they’re allocated 2 for production web use (you may notice the gap, one konked out on us before we could rack it, we’re currently waiting for a replacement), 1 for development, and 1 for file sharing (SAMBA) and staging.
  • 2 CORAID AOE cabinets. 48 drive slots all filled with 1TB hard drives. (that will equate to something like 45 TB of RAIDed storage)
  • 1 CORAID EtherDrive hardware virtualizer. We can slice up those 45TB into logical volumes that can be resized at will.
  • 1 HP Procurve switch for a private SAN network (that’s why it says “NON-ROUTED” on it :) )
  • Public and private inter-box communications handled by our gracious hosts (campus networking services)

In case you’re wondering, we’re going to use the web servers predominately for Plone sites, and the AOE storage for Data.fs storage and windows file sharing of research data. We’re expecting large Data.fs files since we’ll be doing a lot of storage of data directly in Plone too. We may offload to the SAN for file storage in Plone, but we’re not sure just yet. There’s a definite need for some of Plone’s key features for the files in question (versioning, security, etc), and we have to make sure those will work the same with something like FileSystemStorage. We plan to eventually attempt in sillico experiments, crunching the data stored via Plone.

SO, the first ever blitZEN racking day is over, and it was a huge success. Special thanks goes out to ITS for making this possible.

Have you hugged your system administrator today?

This entry was posted in Uncategorized. Bookmark the permalink.

One Response to It’s RACKING DAY!

  1. Just a tip: you’ll want to become familiar with how to read “iostat” output to diagnose poor system performance due to slow IO (on your SAN devices):

    FTR, I’d recommend stress testing performance of the SAN in the ways you might use it from Plone (e.g. as a place where FileStorage files are held) before deploying “for real”.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s