It happened again.
I was notified via a helpdesk change order request that the organization was implementing an RFID solution that would be hosting its data in a SQL Server database. Projections were that this would require approximately 3 TB (that is with a T and that rhymes with C and that stands for Capacity Issues) of space and that we would place this on an existing SQL Server. We could expect approximately 10 transactions a second based upon vendor calculations.
Did you see the part about this going on an existing SQL Server? Did you catch the part where I mentioned I was not notified of this until Monday of this week and the notification occurred through a method that suggested that this was a done-deal? What I did not mention is that they expect(ed) this to be implemented and in place before the end of this month. At the time this gave all teams 28 calendar days to accomplish the task.
A little bit about where the customer and non-DBA technical staff decided where this (these) databases would live… It’s our existing production enterprise cluster already host to 120 separate databases. Things sound a little out of line to you? Well, they certainly did to me.
So once again a DBA is thrown under the bus, and so is the organization, quite frankly, due to not involving the correct personnel from the beginning of the process of vetting, purchasing, architecting, and implementing a technology solution in a large organization. I do applaud the Project Manager though in that I was able to quickly bring this to her attention, and with the aid of all required parties, get the budget amended, and have the project back on track (with a reasonable schedule) within 3 business days and with only 3 short meetings. The databases will now exist on their own 2-node SQL Server 2008 64-bit Enterprise Edition cluster. Net result was an additional six-figure outlay to accomplish the proper architecture (most of the cost associated with 2 processor-based licenses for MS SQL Server.) This is money that should have been allocated from the beginning, and would have been if the correct individuals were included in the process. Trust me, the last thing I enjoy is meetings, but if it saves the IT Department from looking like bufoons because we send people back begging for money then I’m all for it.
This harkens back to a post by Brent Ozar quite recently on his blog asking if DBAs are losing their power. Lately, the only power I seem to have is to tell others that their solutions are incomplete, chinzy, harmful, and not secure.
Of course the questions of a dedicated test/build/train or dev domain has gon unanswered. I suppose they will try to have us place this on our existing SQL 2005 test/dev cluster. Spot anything wrong with that statement?