Just a suggestion...

I don't know what your hardware setup currently is, and I'm not here to blatently rag on you (I'm speak to the SL team here).

I've been a Systems Admin / Engineer for the last 10 years and for the short amount of time I've been part of SL, I can somewhat imagine what kind of hardware requirements you would require for a project such as this, and as I currently maintain the servers for a medical university, along with all of its satellite locations and facilities I can tell you that when I came in, the place was in a constant state of what I like to call "perpetual repair".

Once I insisted on redoing the budget for new hardware and layed out the plans, they finally agreed and we've been humming along fine ever since.

We installed simple rack-mount servers (hey... take your pick. do some research and choose what you like) 500GB RAID hot-swap arrays with a SAN backup system.

That's it. We store / serve over 40 terrabytes of information at any given time. We ended up having to eventually move over to a two T3 connections to handle the network / wireless network loads.

BUT. The point is that after the expense. Once I got the network setup RIGHT, along with all peripherals, etc.... we are now operating with such little downtime and hardware replacement that our budget is already back up to where it was and will be more next year. (200

Just my 2 cents.

I realize it's a big job, and I'm not a paying citizen, so I don't have much clout, but I think you all have a REALLY great thing going here so I hope it works out for the best. I really do.

Take care!
Haplo

Thanks for the good wishes, Haplo. Not being on the ops team, I can't really speak for them, but I think the challenge is less in purchasing hardware, than in managing the complex and extremely intensive transfer of data from resident actions (teleporting, building, moving objects in and out of inventory).

The clusters that were having problems are together known as "the asset server", and carry extremely heavy loads. The good news is that we're adding more hardware to the system which should allow us to distribute the load as well as to provide greater redundancy.