10.29
The following statements and questions are merely random thoughts that came up during postmortem discussions about our efforts to support March Madness traffic. These represent the questions I took with me to HighEdWeb 2010 and will be questions we continually revisit as we plan for March 2011. I just wanted to capture these ideas somewhere I could find them again. Also, by posting them here, I thought perhaps a comment thread may emerge, sparking new ideas.
We are talking about scaling our website, servers, content, etc. Ideally, we need to break the entire puzzle into generic pieces, but how? What is the structure of this particular problem? Content? Servers? How do we define “flexibility” and “scalability?” Is there a way to define the problem, and its pieces, in a generic fashion so that we can see how those pieces fit … regardless of specific technologies? An “agnostic” method of looking at this issue. VMware limits us to scalability. A mainframe would be bigger. What is best way to deal with THIS type of traffic? We need to DEFINE the bottleneck. We don’t have the money to scale everything. Nor do we really WANT to scale EVERYTHING. So, which pieces are PRIORITY for scalability? Internal vs. external? Architectures? And, what about REAL TIME analytics? Are there business types w/ similar business piques/spikes? ANY event could trigger a spike in traffic?
Do we need a more detailed Visio diagram of network architecture to evaluate? Are there infrastructure analysis tools? Open source load analysis tools? How did webSnap affect the overall picture? We had URL rewrite scripts in place that had to of affected processing, etc.
Next Steps:
- Brain dump with Chris (What is REALLY happening on TOAD?) – David/Chris/Tim
- Upgrade TOAD – David/Chris/Tim
- Load balancing tools – Nate
- Walk David through presentation???