Cloud Computing: A Reality Check

Cloud computing is the buzzword in technology circles today. It is sometimes compared with the virtualization of computing power, applications, and storage, thought of as a model to deploy pay-as-you-go web services or perceived to be similar to grid computing. Cloud computing shares characteristics with all of these technology paradigms, yet it has more to offer.

 

Market Opportunity and Consideration Factors

A 2010 IDG survey revealed that-

6% of CIOs polled felt that cost reduction across the board was a critical business priority for the future.

62% felt that optimizing resources and key business processes was going to be a priority.

67% saw improving the marketing time for products and services as critical in the coming years.

A financial services firm that heavily relies on IT-enabled services can benefit from cloud computing, despite the objections mentioned above. Perceived cost savings, ease of scaling-in and scaling-out, faster time-to-market for deploying systems, virtualization of enterprise-wide data as a service, enterprise technology standardization, and the ability to access data and applications on the move are all critical consideration factors that can drive financial services firms to adopt cloud computing.

Current Adoption

Cloud computing has caused more debate than many other recent technological advancements. Concerns around data security and compliance, application availability, and absence of standards have been discussed in a lot of forums. Regardless, there has been a tremendous rise in its adoption by financial services firms over the last couple of years. Some prominent examples include-

NYSE Euronext Capital Markets Community Platform: Recently, NYSE Euronext launched a PaaS community cloud service for the financial services industry, aimed at brokers, dealers, hedge funds, and other market makers. The platform has been set up to host customer applications and services, such as electronic trading, market data analysis, algorithmic testing, and regulatory reporting. The infrastructure consists mainly of storage and virtualization tools from EMC and VMware, running on Xeon-powered blade servers.

NASDAQ OMX Data On-demand: This SaaS cloud service, built with the support of Xignite, provides easy and flexible access to massive amounts of historical level 1 tick data. Its a web application that allows users to purchase data online and access it using an Application Programming Interface (API) or as plain text files.

CME ClearPort OTC Data On-demand: This on-demand SaaS web service is also built on top of the Xignite platform and offers access to the end-of-day OTC settlement, volume and open interest data to support markets available through CME ClearPort.

I-banks Using Cloud for Risk Analysis and Non-core Processes: Now a part of Bank of America, Merrill Lynch used IBM iDataPlex servers as part of an IaaS strategy to build and evaluate risk analysis programs. The servers turn many separate computers into a pool of shared resources, ie, a cloud. Morgan Stanley uses PaaS cloud vendor Force.com for its recruiting applications and has extensive cloud penetration in analytics and strategy.

Gridglo Real-time Energy Apps: The startup, Gridglo, is developing SaaS services to sell energy information to utilities. It is mining energy consumption data from smart meters and combining it with data from other sources, such as real estate, weather, and demographic data, to provide tools for energy forecasting, demand response analytics, and energy credit scoring for categorizing different types of consumers, along with a financial risk energy tool.

Microsoft Azure DataMarket for the Energy Industry: Microsoft DataMarket SaaS cloud services enable the discovery, exploration, and consumption of data from trusted public domains and commercial data sources, such as demographics, health, location based services, real estate, weather, transportation, navigation, etc. It also includes visualizations and analytics to enable insight from that data. All this data can be incorporated into software applications for any device through a common API. Different players in the energy industry are using this platform to create energy forecasting and analytics applications.

Whats Ahead

There are countless opportunities for financial services firms to leverage the benefits of cloud computing by migrating a variety of applications to the cloud. Non-core applications and such business processes as recruiting, billing, and organization-wide travel management can-and should-easily move to the cloud. A number of infrastructure operations, such as data center management, data storage, and disaster recovery should also move to a cloud after a thorough evaluation of different vendors offerings and based on the flexibility of cloud vendors in documenting contracts.

Although very few firms are currently using cloud computing for their core applications, different hosting architectures provided by IaaS cloud providers and new avenues in the community and hybrid cloud space, will drive more firms to move their core applications to the cloud. In fact, core solutions, such as batch processes running throughout the day, analytics, and reporting applications are perfect candidates.

A few scenarios that would be ideal for a cloud deployment include-

Risk Analytics Calculation: Applications that calculate such analytics as cost of trade, current value, yields, Greeks, etc, at the level of a single security, position or portfolio are perfect candidates for a grid based cloud. A cloud based grid service can easily scale up or scale down depending on the data load. Whats more, the applications can be seamlessly deployed on multiple grid nodes, reducing maintenance overhead. Also, since such applications only run for specific durations, dedicated hardware leads to unutilized CPU cycles, which can be optimized by a grid-based cloud. The whole solution can be implemented on a private cloud where existing computing power can be virtualized and made available as an on demand service.

Performance Attribution: Performance attribution provides a framework for examining the relative performance of a fund versus its benchmark. It is a methodology that quantifies the success or added value of an investment strategy. Attribution allows investment managers to identify the factors of the investment process that contributed (positively or negatively) to the performance levels highlighted by performance measurement. Hence, these data-intensive processes need access to a huge amount of historical data for correctly calculating metrics. Performance attribution or benchmark re-balancing applications run at specific times of a day, like the analytics calculation processes. As such, these are ideal candidates to be deployed on a cloud, able to optimize the usage of available computing power and the scale-in and scale-out benefits of an existing grid.

Trade Matching and Reconciliation: A trade matching process gets trade data from multiple brokers and counter-parties and then reconciles it. This process is prone to high volumes during times of peak trading.

The solution is to create a hybrid cloud where the reconciliation process can run on a public cloud for scalability and the data can reside on dedicated database servers in a private cloud. The data from multiple brokers and counterparties can be pushed to the public cloud, which can then be streamed to the private cloud. This can also help avoid creating separate connectivity to new partners and maintaining all those connections simultaneously.

Reference Data Virtualization: Various types of reference data, such as security master data, positions data, holdings and book data, broker and counter-party data, etc, reside in multiple kinds of data sources. These data sources can be internal databases, file systems, or external feeds.

When an application needs to access data from many sources, it can be a challenge to devise strategies that connect those data sources and consolidate and aggregate the data within the application for specific needs. The recommended solution is to build a data virtualization layer that seamlessly federates these different data sources and provides different ways to access the single virtual data source. The layer should be flexible enough to mash up different streams of data according to the requirements of a particular application. Similar to the reference data virtualization layer, a transactional or operational data virtualization layer can be created to support risk management, financial analysis, and compliance reporting. The goal is to make all data available through centralized data services.

Conclusion

Continued growth of cloud computing within the financial services industry will require vendors and firms to overcome its challenges together. The NYSE Euronext community cloud paves the way for these types of collaborative ventures in the future, where multiple firms will have a proportionate stake. And in the areas where data secrecy is more important than collaboration, hybrid clouds with the appropriate allocation of data and applications are recommended.

Cloud computing is a promising paradigm for delivering computing utilities as services. Just as personal computers and servers shook up the world of mainframes and minicomputers, or as smartphones and tablets revolutionized the mobile commerce industry, cloud computing is bringing similar far-reaching changes to the licensing and provisioning of infrastructure and to methodologies for application development, deployment, and delivery.

Some firms have already realized the benefits of cloud computing, which include scalability, cost savings, and time-to-market. Firms that are still looking to leverage the cloud should begin by moving non-revenue generating and non-core systems to the cloud. And, they should consider developing a comprehensive cloud strategy to move core applications to the cloud.