What would a 5 billion share trading day look like?

The current volatility of the financial markets is an example of why understanding demand is so important. The inter-connections across all aspects of our global village, Insurance, Capital Markets, commodity supplies such as energy, forces questions about the characteristics of demand to be answered. Example: How would a financial services company answer "What does a 5 billion share day look like?" How would it affect all of your infrastructure throughout the day.

For an Financial Services company, it is not just about the sheer volume, it is about increased errors rates in the incoming data, increased alerts, increases in overnight processing which already have tight operational windows. It is about increased network latency as temporary servers are brought on line, anywhere there is room to handle the demand.

Even if your company is not financial services, the cascading impact of any large event on transaction traffic and queries, are important for IT executives to consider. In any dynamic market, financial services or otherwise, the demand profile would change as the volume increases. Example- after Thanksgiving, retail stores can expect surges in volume that will last through the New Year, but what is the proportion of browsing to sales, and how does that affect fulfillment. With the increase in streaming media as a key attractor for potential buyers, what is the effect on the network, storage and compute? How does meeting the needs of the client facing systems affect the resources required for back end and fulfillment systems? How does severe weather affect a whole range of companies? The needs could change inter-day, (heavy client facing demand during the day and heavy fulfillment and inventory replacement at night).

Notice how all the questions just posed reflect temporary conditions, where the need for extra resources drops dramatically after a certain period. A company's infrastructure that is organized by supply based logistics (e.g. DB and application servers arranged by types in different rows for easier maintenance), will not be agile enough when the need is there. Highly collaborative applications will not necessarily be placed in optimal positions.

Understanding the characteristics of demand can be difficult, but so is reacting to a different need several times a day, while never satisfying anyone, and "blowing the budget" in the process. It starts with modeling the business and understanding how business transactions drive all other behavior. But once modeled, IT and the business can talk about growth in productive ways.

It means having comprehensive instrumentation that can be related to business trends, including volume induced latency and trends that indicate how fast demand will rise.

As has been stated in this blog series, the network has analogies to the human body, and it is just as important to remember that the network is the digital analog of the business. The human body needs to be checked when under stress, so does the infrastructure and for the same reasons. Stress can make you sick.