Enterprise IT as a Marketplace- Infrastructure as a Service

Now that the gold rush of mortgage backed securities trading has led to a severe crash, what do surviving investment banks do with all that hardware (i.e.10 million dollars plus for each institution or more)? Why not become a service marketplace, reduce the risk that comes from directly trading in such securities, and get a cut of the proceeds by being the facilitator of flow? Banks should become flexible and offer their technical expertise as a service.

So why should banks become an infrastructure service provider for hedge funds?:

1-No use in sight-Banks may not be able to generate the kinds of business that would justify running that large store of hardware that used to run the derivative trading system infrastructure.

2 -Wasted opportunity-No one is going to buy the used hardware that ran the derivative trading systems anytime soon.

3-History-After previous significant crashes (railroad crash of the 1880s, Internet crash of 2000), significant infrastructure improvements were the legacy which provided a leg up to the new opportunities, bolstering the economy on the rebound. There is an opportunity to utilize these unused/ underutilized resources in a similar fashion.:

4-Change the game and win-Banks would do well to create a real discipline of infrastructure service provider; it will make their own IT a better strategic partner, building up the next competitive landscape.

This argument was made in a recent article about the new role of banks in the market crash aftermath:

http://weblog.infoworld.com/real-time-enterprise/archives/2008/11/wall_street_as.html?source=rss

The opportunity is that banks can supply the necessary robust infrastructure, providing a service to many hedge funds that will take over the risky trading. These hedge funds were rarely inclined to build that level of robustness into their own infrastructure. That article concluded that providing Infrastructure as a Service (IaaS) is not easy,

Migrating from an informal, get it done; nail-the-support-guys-foot-to-the-trading-desk-kind of mentality does not scale and only works when you are making money hand over fist. To provide IaaS, whether for trading or any other business function, means that service level contracts need to be defined in business terms. For example, client X will lose 1 million dollars / hour when their structured product trading desk doesn't have access to the liquidity pool. Building such a robust infrastructure from scratch is not only expensive, but requires behavior change. It is a paradigm shift.

The kinds of services that these banks would be expected to supply would all be in support of trade execution services:

• Data federation and information integration
• Messaging, including linkages to liquidity providers
• Distributed transactional caching

There are emerging tools and products that when integrated with the current existing environments would provide a migration toward such a service, the following are just a few of the functions.

• Wire speed message transformation
• Protocol switching
• Dynamic, demand-driven real-time allocation of compute resources
• Solid state non-spinning disks
• Dense compute platforms that save datacenter space and cooling energy, ensuring lower latency
• Wire speed fine grained security authorization that supports third-party authorization (SAML)
• Network attached processors that provide all of the benefits of a large machine in terms of memory and CPU cores, but little of the drawbacks

All of these emerging products would need to be integrated and hardened; many would help save current investments in legacy technology until the service business model could be proven.

None of it will work without a significant cultural shift on how investment banks run their operation. People, Process and Technology shifts are all required. Process and People will be the focus of the next blog; Technology will get further attention in subsequent blogs.

Sheppard & Tony

• Data Center
• Software
• Infrastructure Service